Advances in Child Development and Behavior Volume 22

ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR

Volume 22

Contributors to This Volume Margarita Azmitia

Leila Regina de Paula Nunes

Judy S . DeLoache

Marion Perlmutter

Roger M. Downs

Claire L. Poulson

David Estes

Daniel W. Smothergill

Alan G . Kraut

Steven F. Warren

Lynn S. Liben

Henry M. Wellman

Nora Newcombe

Jacqueline D. Woolley

ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR

edited by Hayne W. Reese Department of Psychology West Virginia University Morgantown, West Virginiu

Volume 22

ACADEMIC PRESS, INC. Harcourt Brace Jovanovich, Publishers

San Diego New York Berkeley Boston London Sydney Tokyo Toronto

This book is printed on acid-free paper. @

COPYRIGHT 0 1989 BY ACADEMIC PRESS, INC. All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means. electronic or mechanical. including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher.

ACADEMIC PRESS, INC. San Diego. California 92101 Uiiited Kir7,qdonr Edition puhlishrtl hji ACADEMIC PRESS LIMITED 24-28 Oval Road. London NW I 7DX

LIBRARY OF CONGRESS CATALOG CARD NUMBER:

ISBN

0-12-009722-2 (alk. paper)

PRINTED IN THE llNlTED STATES OF AMERICA XY

yo

YI

92

9

x

7

6

s

4

3

z

I

63-23237

Contents

Contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ix

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

xi

The Development of Representation in Young Children JUDY S . DELOACHE Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Young Children‘s Memory lor the Location of :I Hidden Object . . . . . . . . . . . . . . . . Rapid Change in Young Children’s Reprewiltntiunal Functioning . . . . . . . . . . . . . . . . Spatial Cognition: Imitation 01 Object Placement acres\ Spaces . . . . . . . . . . . . . . . . . VI Analogical Reasoning ill Young Children . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII Symbolization in Young Children . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI11 . Related Research on Multiple Reprewntations . . . . . . . . . . . . . . . . . . . . . . . . . 1x Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 11 111 IV V

2 2 3 4 13 IS

75 37 35 36

Children’s Understanding of Mental Phenomena

.

.

DAVID ESTES HENRY M . WELLMAN AND JACQUELINE D . WOOLLEY I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II . Childhood Realism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 . Mental Entitie5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I V. PriorStudieh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. CurrentStudies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . General Diwusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

V

41

43 35 46 4X X2 XS 86

vi

Contivits

Social Influences on Children's Cognition: State of the Art and Future Directions MARGARITA AZMITIA AND MARION PERLMUTTER I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Theoretical Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Evidence for the Impact of Social Agents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV . A Framework for Considering Developmental Change in Social Influences on Cognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Additional Theoretical and Methodological Issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

90 92 103

124 131 135

136

Understanding Maps as Symbols: The Development of Map Concepts in Children LYNN S . LlBEN AND ROGER M . DOWNS 1. Why Maps? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . I11 . Where Have We Been'? Review of Past Mapping Research . . . . . . . . . . . . . . . . . . . . . . IV. The Mapping Project at Penn State (MAPPS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . V. Map Understanding in Young Children Revisited . . . . . . . . . . . . . . . . . . . . ... VI . Maps as Symbolic Representations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

146 147 151)

176 190 193

196 198

The Development of Spatial Perspective Taking NORA NEWCOMBE I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. What Does Perspective Taking Assess? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. How Can Spatial Location Be Encoded'? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Factors Affecting Success on Perspective-Taking Tasks. . . . . . . . . . . . . . . . . . . . . . . . . V. RelatedTasks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Conclusion . . . . . . . . . .............................................. References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

203 20.5 212 217 236 240 241

Developmental Studies of Alertness and Encoding Effects of Stimulus Repetition DANIEL W. SMOTHERGILL AND ALAN C . KRAUT I . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11. The Stimulus FdmilIarizarion Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111. Components of Attention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . IV. Age. Stimulus Familiarization Effects. and Latent Inhibition . . . . . . . . . . . . . . . . . . . .

250 251 254 256

V . Mechanisms of Familiarization Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VI . Stimulus Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII . Fainiliarimtion as a Tool in Studying Reading Acquisition . . . . . . . . . . . . . . . . . . . . . . VIII . Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

258 260 2b2

265 26X

Imitation in Infancy: A Critical Review

.

I. II. 111 .

IV V

VI . VII . VlII .

.

CLAIRE L . POULSON LEILA REGINA DE PAULA NUNES AND STEVEN F. WARREN ... Introduction Ewly Conceptualizations of Infant Imitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ... I h d o p m e n t n l Trends in Diverse Imitative Responses . . . . . . . . . . . . . . . . . . . . . . . . . Variables Aflecting Inlant Iniitative Performance . . . . . . . . . . . . . . . . . . . . . . . . . . . Imitation and Mother-Infant Interaction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods (11 Evoking Imitation i n Infants . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Neonatal Imitation of Nonvisible Actions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Past and Future Trends in the Study of Infant Imitation . . . . . . . . . . . . . . . . . . . . . . Kclcrence\ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

272 214 215 2x1 2x4

285 288 293 29s

Authorlndex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

299

Suh.jcct Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

309

Content\ oi Previous Volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

321

This Page Intentionally Left Blank

Contributors Nuinbm in parentheses indiciile thc pages on which the author\' contributions begin

MARGARITA AZMITlA

Dcpcrrtinent of Psychology, Florida lnternntional Universityp Miami, Florida 33199 (89) JUDY S . DELOACHE

Depurtinrnt of Humtrn Development and Fumily Ecology, University of' Illinois, Urhanu, Illinois 6/80] ( I ) ROGER M. DOWNS Deimrtinent of' Ceogruphy, Tht. Pennsyhaniu Sttire University, University Purk, Pennsylvuniu I6802 (145) DAVID ESTES

Department of Hiitnun DevPlopment r i n d Family Studies, The Pennsylvuiiici Stcite University, Universir~~ Pcrrk, Pt~nnsylvuniaI6802 (41) ALAN G. KRAUT

Aniericun Psychologiiul Society, Washington D. C . 20003 (249) LYNN S . LlBEN

Depurtrnent of Psychology, The Pennsylvunirr Statc University, University Purk, Penns$vuniu 16802 ( 145) NORA NEWCOMBE

Department qj' Psychologj: Ternple Uniwrsity, Philadelphiu, Peiinsylvaniu 19122 (203) LEILA REGINA DE PAULA NUNES

Depiirtuinento de Psicologiu, Univc~rsiclackFederul de SLIo Curlos, SLIo Curlos 13560, Sao Puulo, Brazil (271) MARION PERLMUTTER

Department of Psychology m r l Institute of' Gerontology, Unitvrsity oj' Michigun, Ann Arbor, M i c h i p n 48/09 (89) CLAIRE L. POULSON Queens College cind the Grricluate School, City University oj'New York, Flushing, New York 11367 (271) DANIEL W. SMOTHERGILL Departrnent of Psychology, Syracuse University, Syracuse, New York I3244 (249)

1x

Conlributors

X

STEVEN F. WARREN

Department of Special Education, George Peabody College, Vanderbilt University, Nashville, Tennessee 37203 (271) HENRY M. WELLMAN

Department of Psychology, University of Michigan, Ann Arbor, Michigun 48109 (41) JACQUELINE D. WOOLLEY

Department of Psychology, University of Michigan, Ann Arbor, Michigun 48109 (41)

Preface The amount of research and theoretical discussion in the field of child development and behavior is so vast that researchers, instructors. and students are confronted with a formidable task in keeping abreast of new developments within their areas of specialization through the use of primary sources, as well as being knowledgeable in areas peripheral to their primary focus of interest. Moreover, journal space is often simply too limited to permit publication of more speculative kinds of analyses that might spark expanded interest in a problem area or stimulate new modes of attack on a problem. The serial publication Advuttces it1 Child Development and Behavior is intended to ease the burden by providing scholarly technical articles serving as reference material and by providing a place for publication of scholarly speculation. In these documented critical reviews, recent advances in the field are summarized and integrated. complexities are exposed, and fresh viewpoints are offered. They should be useful not only to the expert in the area but also to the general reader. No attempt is made to organize each volume around a particular theme or topic, nor is the series intended to reflect the development of new fads. Manuscripts are solicited from investigators conducting programmatic work on problems of current and significant interest. The editor often encourages the preparation of critical syntheses dealing intensively with topics of relatively narrow scope but of considerable potential interest to the scientific community. Contributors are encouraged to criticize, integrate, and stimulate, but always within a framework of high scholarship. Although publication in the volumes is ordinarily by invitation, unsolicited manuscripts will be accepted for review if submitted first in outline form to the editor. All papers-whether invited or submitted-receive careful editorial scrutiny. lnvited papers are automatically accepted for publication in principle but may require revision before final acceptance. Submitted papers receive the same treatment except that they are not automatically accepted for publication in principle and may be rejected. The Advunces series is usually not a suitable place of publication for reports of a single study, or a short series of studies. even if the report is necessarily long because of the nature of the research. The use of sexist

xi

xii

Prefoce

language, such as “he” or “she” as the general singular pronoun, is not acceptable in contributions to the Advances series. The use of “he or she” (or the like) is acceptable. I wish to acknowledge with gratitude the aid of my home institution, West Virginia University, which generously provided time and facilities for the preparation of this volume, I also wish to thank Dr. Lynn S . Liben and Dr. Lewis P. Lipsitt for their editorial assistance, and Mrs. Ann Davis for her excellent secretarial services. Hayne W. Reese

THE DEVELOPMENT OF REPRESENTATION IN YOUNG CHILDREN

Judy S. Debache DEPARTMENT OF HUMAN DEVELOPMENT AND FAMILY ECOLOGY UNIVERSITY O F ILLINOIS URBANA, ILLINOIS 61801

1. INTRODUCTION 11. REPRESENTATION 111. YOUNG CHILDREN’S MEMORY FOR THE LOCATION OF A HIDDEN OBJECT IV. RAPID CHANGE IN YOUNG CHILDREN’S REPRESENTATIONAL FUNCTIONING A. STUDY 1 B. STUDY 2: INTERMEDIATE AGE GROUP V. SPATIAL COGNITION: IMITATION OF OBJECT PLACEMENT ACROSS SPACES

VI. ANALOGICAL REASONING IN YOUNG CHILDREN A. STUDY 3: SIMILARITY OF OVERALL SPACE VERSUS INDIVIDUAL PLACES B. STUDY 4: IDENTICAL SPACES VII. SYMBOLIZATION IN YOUNG CHILDREN A. YOUNG CHILDREN’S UNDERSTANDING OF MAPS B. MODEL AS SYMBOL: THE DUAL REPRESENTATION HYPOTHESIS C. STUDY 5 : YOUNG CHILDREN’S USE OF PHmOGRAPHS VERSUS MODEL D. SYMBOLIC PLAY VIII. RELATED RESEARCH ON MULTIPLE REPRESENTATIONS IX. SUMMARY AND CONCLUSION REFERENCES

1 ADVANCES I F i CHILD DEVELOPMtNT A N D BEHAVIOR. VOL. 22

Copyright 6 1989 by Academic Press, Inc. All rights of reproduction in any form reserved.

2

Judy S Debache

I. Introduction A young child watches as a toy is hidden somewhere in a room. Then the child is shown a scale model of the room and told that a miniature toy is hidden in the “same place” in it. A 3-year-old knows precisely where to search for the hidden miniature toy, but a 2.5-year-old child is mystified. The phenomenon that is the primary focus of this article is a very rapid change that occurs between 2.5 and 3 years of age in children’s appreciation of the correspondence between two spaces and their understanding of the implications of that correspondence. The 3-year-old understands both that the model represents the large-scale room and that his or her memory representation of the original hiding event in the room can be used to figure out where the miniature toy is hidden in the model. In this article I report a series of studies of this phenomenon and consider it in the context of related domains, including spatial cognition, analogical reasoning, and, especially, mental representation, both in terms of memory for an event and symbolic representation.

11. Representation The topic of representation lies at the heart of cognitive psychology. A cognitive or internal representation refers to a stored system of information that preserves some, but not all, of the information in the world it represents. As Palmer (1978) put it: “The nature of the representation is that there exists a correspondence (mapping) from objects in the represented world to objects in the representing world such that at least some relations in the represented world are structurally preserved in the representing world” (p. 266). Within developmental psychology, representation has been of particular concern, in part because of the central role assigned to changes in representational capabilities in major developmental theories. Piaget, Vygotsky, and Bruner all posited that qualitatively different modes of representation underlie different stages in cognitive development. More recent investigators (e.g., Kosslyn, 1978; Presson & Somerville, 1985) have concentrated more on developmental changes in representational functioning and less on changes with age in the underlying form or mode of representation. The present article follows this approach and is concerned with the early development of representational functioning, with changes in young children’s ability to use their memory representation of an event to reason about a different event. Mandler (1983) drew a distinction between two different senses of the term representation, and this distinction is especially useful for organizing this article. The first sense-“representation in the broad sense”-is roughly

Representation

3

equivalent to knowledge. It includes “both knowledge and how that knowledge is structured’’ (p. 420). The second-the “narrow sense”-refers to symbolic representation: “words, artifacts, or other symbolic productions that people use to represent (to stand for, to refer to) some aspect of the world or some aspect of their knowledge of the world” (p. 420). Although both senses of representation have been the focus of a great deal of developmental research, the two have been investigated almost completely independently by different sets of researchers studying quite different kinds of phenomena. The broad sense of representation has been studied by investigators who are interested in the structure of knowledge and memory. Examples include research on the acquisition of scripts about everyday events (Nelson & Gruendel, 1981), the development of story grammar structure (Mandler & Johnson, 1977), the influence of knowledge structures on memory (Chi, 1978), and the development of spatial cognition (Liben, Patterson, & Newcombe, 1980). The development of symbolic representation has been studied in a variety of media, including the production and understanding of metaphors (Gentner, 1978; Winner, 1979), notational systems (Cohen, 1985; Karmiloff-Smith, 1979), maps (Bluestein & Acredolo, 1979; Presson, 1982), and drawing (Freeman, 1980). This distinction between two senses of representation is useful for organizing the large body of research on representational development, and it brings into focus many real and meaningful differences between the two. At the same time, however, the distinction is somewhat arbitrary, and serious thought about the relation between knowledge and symbolic representation might prove illuminating. This article concerns both senses of representation; some of the studies to be summarized are primarily involved with representation in the broad sense and some mainly concerned with the narrow sense. Ultimately, the interconnectedness of the two is highlighted.

111. Young Children’s Memory for the Location

of a Hidden Object The research that is described here grew out of work that I have been doing for several years on the early development of memory. My colleagues and I have conducted numerous studies investigating the ability of children between 18 and 30 months of age to remember the location of an object hidden in a large-scale environment-a room ( D e b a c h e & Brown, 1979, 1983, 1984; Debache, Cassidy, & Brown, 1985). The memory task is presented to the child as a game of hide-and-seek to be played with a small stuffed animal (Big Bird, Snoopy). The child is told that Snoopy is going to hide and that he or she should remember where

4

Judy S. Debache

Snoopy is in order to find him later. The child then watches as the toy is hidden in some natural location in the room. The hiding locations consist of places common to almost any living room, such as under a couch or chair cushion, behind a door, inside a cupboard, and so forth. A timer is set for a specified interval, and the child is told that when the bell rings, he or she can find the toy. Young children readily understand the goal and rules of this game, and most of them greatly enjoy playing it. Our 1.5- to 2.5-year-old subjects have exhibited excellent memory in this task. The level of performance across studies is quite high, averaging over 80% errorless retrievals. (An errorless retrieval is defined as the child searchingfirst at the correct location.) This high level of performance in the hide-and-seek game has proven to be extremely robust. Within the age range of children tested, age has not been consistently related to memory performance in the basic hide-and-seek task. In some studies, we have found significant age differences, with older subjects (24- to 30-month olds) outperforming younger ones (18- to 24-month-olds). In other experiments, no age effects have appeared. Thus, a small degree of developmental change in memory for spatial location may occur over this age period, but even 1.5-year-olds generally perform quite competently in the basic task. The amount of practice or experience with the game also appears to be unimportant; children given pretraining do no better than children participating in the game for the first time, and performance generally remains stable over days. Whether the children are instructed to remember the toy’s location or not is irrelevant. The length of the delay intervals (from a few minutes to several hours or overnight) has little effect on performance. The familiarity of the space in which the game is played is not important; children tested in an unfamiliar laboratory playroom achieve the same high level of success as children searching in their own living rooms. The robustness of early memory in this task led me to ask the question underlying the research summarized in this chapter: Given the existence of an accurate and accessible memory representation, to what extent are young children capable of using that representation, of transforming or translating the original representation for application to a new problem? More generally, this question is about flexibility of thought-being able to take something that one knows, that one has learned in one situation, and apply it in a new and different context.

IV. Rapid Change in Young Children’s Representational Functioning The original experiment that was designed to address these issues involved a modification of the hide-and-seek task, a modification that resulted in major

Representation

5

effects on young children’s performance. Instead of retrieving the toy that he or she had observed being hidden, the child was asked to retrieve an analogous toy that was concealed in an analogous location in a different space. More precisely, the child watched as the experimenter hid a toy-a stuffed dog-in a location of the type typically used in the hide-and-seek task-say, behind the chair in a room. Then, the child was asked to find a different, miniature dog that had been hidden without the child’s observing behind a miniature chair in a scale model of the room. This task thus required the child to draw on his or her memory representation of the original hiding event (toy dog being placed behind the chair) to figure out where a different toy dog could be found. The child had to translate his or her memory representation for use in a different context. A.

STUDY 1

The first step in Study 1 (DeLoache, 1987, Experiment 1) was to construct a scale model of my laboratory. Figure 1 shows the layout of the laboratory playroom and a small adjoining room in which the model was situated, aligned in the same orientation as the room. The lab is furnished like a standard (albeit rather shabby) living room. It is carpeted and contains a couch, a coffee table, a large armchair, a small (child-size) wooden dresser, large floor pillow, small throw pillows on the couch, and a built-in set of bookcases and cupboards along one wall. The plywood model of the laboratory room was constructed to approximate scale. It was open at the top and on one side (the wall opposite the couch and chair), to facilitate the children’s access to it. The model’s features and furnishings duplicated those of the room, including carpeting, the built-in wall unit, and miniature items of furniture corresponding to those in the room. Some of the model furnishings were made as perceptually similar as possible to their counterparts (except for size): The miniature couch was covered in the same fabric as the full-sized couch, and the miniature dresser was constructed of unstained, light wood like the dresser in the room. Other items were deliberately constructed to share less perceptual similarity with their counterparts. The armchair in the model was made of two blue fabric pillows resting on a wooden frame, but the large armchair was a remarkably ugly black vinyl recliner. The floor pillows in the two spaces were of different colored fabric. 1. Procedure

The experimental procedure was originally designed to make sure that all subjects understood the correspondence between the model and the room. Each experimental session began with a fairly extensive familiarization phase during which the experimenter explicitly pointed out the correspondence. She

6

Judy S. &Loache

Table

Floor Pillow

Fig. I. Luyout of the experimentalspace Darkened a m in the xale model cornpond to labeled items of furniture in the mom. (Room dimensions 4.80 x 3.98 x 2.54 m. Model dimensions: 71.1 x 64.3 x 33.5 em.1

began by presenting the two toys that would be hidden, introducing one as “Daddy Snoopy” and the other as “Baby Snoopy.” She then oriented the child to the room and to the model in the adjacent room, explaining, “This is Daddy Snoopy’s big room, and this is Baby Snoopy’s little room. Look-their rooms are just alike” The experimenter then demonstrated the correspondencebetween the pieces of furniture in the two rooms. She carried each item of furniture from the model into the room and held it up to its counterpart: “Look-this is Daddy Snoopy’s couch, and this is Baby Snoopy’s little couch.”

7

Representation

In the next part of the familiarization phase, the child was asked to imitate the experimenter’s placement of the toy. For the placement trials, the experimenter explained to the subject that “Daddy and Baby Snoopy like to do the same things. When Daddy Snoopy sits on his chair [placing the large dog on the chair in the room], Baby Snoopy likes to sit on his chair, too [placing the little toy on the chair in the model].” After demonstrating two such placements, the experimenter put the large toy dog on a piece of furniture in the room and tried to get the subject to place the small dog in the corresponding location in the model. The last part of the familiarization phase was a practice trial similar to the following experimental trials, with the difference that, if necessary, the child was prompted with the name of the hiding place: “Remember, Daddy Snoopy is in the cupboard, so where’s Baby Snoopy?” The experimenter never labeled the hiding places during the subsequent experimental trials. In the familiarization phase, as well as during the subsequent trials, space was counterbalanced. The experimenter placed and hid the toy in the room for half the subjects and in the model for the other half. For ease of presentation, I continue to describe the case in which the experimenter hid the toy in the room. Immediately following the familiarization phase, each child received four experimental trials, each of which involved three parts, as shown in Table I. a. Hiding Event. The child watched as the experimenter hid one of the toys in one of the spaces. The experimenter always called the child’s attention to the act of hiding, but she never referred to the hiding place by name: “Look, Daddy Snoopy is going to hide here.” The child was told that the TABLE I Procedure for Each Experimental Trial Phase 1. Hiding event

Event

E hides Toy 1 in Space I and tells S that E2 has hidden Toy 2 in the same place in Space 2.

2. Retrieval Ianalogous object

3. Retrieval 2original object

S is taken from Space 1 to Space 2 and asked to retrieve Toy 2. S

Instructions

“Look, Daddy [Baby] Snoopy is going to hide here. [EZ] is going to hide Baby [Daddy] Snoopy in the same place in his room.”

is reminded of the correspondence between the hiding places.

“Can you find Baby [Daddy] Snoopy? Remember, he’s hiding in the same place as Daddy [Baby] Snoopy.”

S returns to Space 1 and is asked to retrieve Toy 1, the toy the S

“Now find Daddy [Baby] Snoopy.”

had observed being hidden.

8

Judy S. DeLoache

second experimenter (who had previously been introduced) would hide the other toy: “Now Mary is going to hide Baby Snoopy in the same place in his little room.” When the second experimenter announced that “Baby Snoopy is hiding,” the child was led into the adjoining room-without having retrieved the toy that he or she had watched being hidden.

b. Retrieval I-Analogous Object. The child was asked to retrieve the analogous toy. In our example, having watched the experimenter hide Daddy Snoopy in the room, the child was now asked to retrieve Baby Snoopy. On every trial, before permitting him or her to search, the experimenter reminded the child of the correspondence between the two hiding events: “Remember, Baby Snoopy is hiding in the same place as Daddy Snoopy.” If the toy was not found on the child’s first search, he or she was allowed to continue searching other locations. If the child stopped searching at any point, the experimenter provided a series of prompts or hints. The first prompt was simply to remind the child again that the toy was in the “same place” as the other one. If this failed to elicit a correct search, the experimenter provided more explicit hints until the child found the object (the point being to maintain the child’s motivation for the task). After the child retrieved the toy, either on the first or a prompted search, he or she was taken back to the original space. c. Retrieval 2 - Original Object. Next, the child was asked to retrieve the original toy that he or she had observed being hidden at the beginning of the trial-Daddy Snoopy in our example. Retrieval 2 was thus a standard memory trial. The child was again permitted to continue searching if his or her first search was incorrect, and the trial always concluded with the child’s retrieving the toy. The second retrieval served as a memory check and was crucial for interpreting the children’s behavior on Retrieval 1. Failure to find the analogous toy on Retrieval 1 could either reflect difficulty translating the memory representation of the original hiding event or the absence of a representation the child might have simply forgotten where the original toy was. If, however, the child could still retrieve the original toy after failing to find the analogous one, the problem could not be with the child’s memory representation of the original event, but must lie with the ability to use or translate that representation. The subjects for the first study were 32 children, half girls and half boys. There were 16 subjects in a younger group (30-32 months), with a mean age of 31 months, and 16 in an older group (36-39 months), with a mean age of 38 months. The sample for this and all the subsequent studies was

9

Representation

predominately middle class and white. As mentioned before, the space in which the toy was originally hidden was counterbalanced: Half the males and females in each age group were assigned to the Hide in Room-Retrieve in Model condition, and half to Hide in Model-Retrieve in Room. Controlling whether the toy was hidden first in the room or in the model was important, because some investigators have warned that processes and behavior may differ for large- and small-scale spaces (e.g., Acredolo, 1977). For example, Siegel, Herman, Allen, and Kirasic (1979) found an “asymmetrical translation effect”: Children’s reconstruction from memory of a model town was less accurate when they were required to perform a translation from a small-scale to a large-scale space than when they went from a large- to a small-scale space. The same four hiding places were used for all subjects: couch, dresser, floor pillow, and chair. Thus, each subject saw the toy hidden twice in high-similarity places (the couch and dresser, which had been made as physically similar as possible) and twice in low-similarity hiding places. Two orders of hiding places (counterbalanced with age, space, and gender) were constructed so that there was no regular pattern of hiding places over trials, and each block of two trials had one high- and one low-similarity hiding place.

2. Results The results were dramatic. Figure 2 shows the percentage errorless retrievals of the analogous toy (Retrieval 1) and of the original memory object (Retrieval 2). The interaction of Age x Retrieval 1 versus Retrieval 2 was highly significant.

Older Younger

8

0

Retrieval 1 (Analogous Location)

Retrieval 2 (Original Location)

Fig. 2. Percenrage errorless retrievals achieved by the two age groups in Study 1.

10

Judy S. DeLoache

Both age groups clearly knew where the original object was hidden (Retrieval 2). The level of performance was exactly as expected from previous research with the basic hide-and-seek task (Debache & Brown, 1979, 1983, 1984; Debache et a/., 1985). Based on that knowledge, the older children also knew where to look for the analogous object (Retrieval 1). In other words, they used their memory representation of the location of one hidden object to draw an inference about where a different object could be found. They were so adept at making this inference that their success in finding the object they had witnessed being put in its hiding place was virtually identical to their success at retrieving the toy they had not seen being hidden. Undoubtedly, these children understood the relation between the two spaces. Unlike the older children, the younger ones did not draw an appropriate inference from their knowledge of where the original object was hidden. The same task that was extremely easy for the older group was extremely difficult for the younger one; in fact, they simply could not do it. Tivo variables that we had thought might influence performance did not. No difference at all was found as a function of small- versus large-scale space. Performance in the Hide in Model-Retrieve in Room and Hide in RoomRetrieve in Model conditions was equal; thus, we obtained no asymmetrical effect such as that reported by Siege1 et d. (1979) in the children’s translation between the two spaces. Also, the physical similarity of the individual hiding places had no effect for either age group; errorless retrievals were equal for high- and low-similarity hiding places. The most striking aspect of the results of the first study is that so large a difference in performance occurred with a relatively small difference in age. Examination of the individual scores revealed very few deviations from the overall pattern shown in Fig. 2. Among the younger subjects, only two children achieved as many as two correct retrievals; none had three or four, and many never found the analogous toy in the first place they searched. In contrast, only one of the older subjects received a score lower than two. B. STUDY 2: INTERMEDIATE AGE GROUP

Study 1 produced little evidence of intermediate levels of performance. Seeking to understand better the nature of the dramatic difference between the older and the younger groups, 1 decided to examine a group of children of intermediate age. The main point of Study 2 was to test children whose age was between those of the older and younger groups in the first experiment to see if their performance revealed a developmental level in between the other two. Accordingly, for the second study, 16 children (5 males and 11 females) between the ages of 33 and 35 months (M = 34 months) were tested in the model task.

Representation

I1

As expected, the intermediate age subjects’ performance (83%) on the memory check (Retrieval 2) was equivalent to that of both age groups in Study 1. With respect to the retrieval of the analogous object (Retrieval I), mean performance was in between that of the other two age groups. The intermediate age group retrieved the hidden toy on 38% of their Retrieval 1 trials; this was somewhat better than the 15% of the younger subjects in Study I, but it was significantly below the 77% achieved by the older children. This intermediate level of performance could occur in at least two ways, however. It could reflect gradual progress in children’s mastery of the model task, with most children showing a steadily increasing score with age. Alternatively, the group mean could reflect very good performance by a few individuals and very poor performance by the rest. The Retrieval 1 data for individual children support the latter alternative. The intermediate level of Retrieval 1 performance resulted from the fact that four intermediate-age children, versus none of the younger ones, had 75% or 100% errorless retrievals. Perhaps, however, a gradual increase in children’s understanding of the task might not show up in their retrieval performance but reveal itself in other aspects of their behavior instead. To examine this possibility, we considered the children’s response to prompts and their performance on the imitation (familiarization) trials. On the basis of this information, and independently of their number of correct retrievals, the 16 subjects in Study 2 were classified into two groups-one group that clearly understood the correspondence between the spaces and a second group that clearly did not. The first group, “correspondence” (6 children; mean age 33.8 months), gave good evidence of understanding the correspondence between the room and the model. These children usually benefited from the prompt when they had failed to search correctly; that is, after the child searched one wrong place, the experimenter’s comment, “Remember, Baby Snoopy is hiding in the same place as Daddy Snoopy,” was sufficient to elicit a correct search. This group also performed correctly on at least some of the familiarization trials in which they were asked to imitate the experimenter’s placement of the toy. Informal evidence also attested to these subjects’ comprehension of the correspondence between the two spaces. During the familiarization phase, some of them actually anticipated the experimenter’s explanation of the relation between the room and the model: One child spontaneously commented, “They’re the same,” and another started explaining the relationship to the experimenter. In addition, some of the children in this group displayed a very interesting and revealing strategy. Occasionally, during Retrieval 1, a child who appeared puzzled as to the correct location would return to the original space and scan it, visually seeking out the hiding place, and then go immediately back and retrieve the toy in the analogous space. One boy who

Judy S. DeLoache

12

had immediately grasped the correspondence on his own before the experimenter explained it, said, “Let’s check,” after failing to find the toy on his first Retrieval 1 search on Trial 1. He then used this checking or reminding strategy on every trial before doing any Retrieval 1 searches-and he was always correct. This behavior reveals an explicit awareness of the correspondence between the two spaces. The other group, “noncorrespondence” (10 children; mean age 33.5 months), gave no clear evidence of understanding the correspondence bet ween the two spaces. They were almost never correct on the imitation trials, and they almost never profited from the prompts. Figure 3 shows the retrieval data for these two groups of subjects. The data for the correspondence group look just like those for the older children in the previous experiment, and the noncorrespondence group’s data closely resemble the younger subjects’. Thus, the results of Study 2 offer no evidence of gradually increasing competence. Indeed, they show that the relevant abilities emerge over an even shorter time frame than was indicated by Study 1, with the transition from incomprehension to mastery taking place in a period of only 3 or 4 months. Most developmental phenomena take place more gradually, with identifiable steps or stages of improvement. The main questions that emerge from the initial two experiments are (1) What is responsible for the younger children’s failure in the model task? and (2) Why is development so abrupt? In the following sections, I consider these questions in terms of three different kinds of cognitive skills relevant to this task-spatial cognition, analogical reasoning, and symbolic representation.

Correspondence

+ Non-Correspondence 220 W

$ 0

Retrieval 1 IAnalogous Location)

Retrieval 2 loriginal Location)

Fig. 3. Percentage errorless retrievals achieved by the two groups of intermediate-age subjects in Study 2.

Representation

13

V. Spatial Cognition: Imitation of Object Placement across Spaces The model task is in part a spatial cognition problem, one that requires space-to-space reasoning. Various elements of the problem include (1) understanding that two spaces are related to each other, (2) mapping the individual places (or objects) within one space onto the corresponding places within the second space, and (3) mapping the spatial relations among the places or objects in one space onto the corresponding spatial relations in the second. The third element, mapping spatial relations, may or may not be necessary for success in the model task, but the first two are presumably required. Very little research exists on the early development of space-to-space reasoning or the understanding of scale models of large-scale spaces, although some studies have been done with 3- to 5-year-old children (Blades & Spencer, 1986). One relevant body of research that involves the extrapolation of information from one real space to another is the work on perspective taking, or mental rotation. A large number of studies, based on Piaget’s views on preoperational thought (Piaget & Inhelder, 1956), have addressed the development of the ability to imagine how a spatial display would appear from a perspective different from the child’s own. In several of these studies, two identical smallscale spaces have been used. The experimenter places an object in one space and asks the child to place an identical object in the same place in the other space. The conditions that are of most interest in terms of perspective taking are those in which the child’s space is rotated relative to the experimenter’s space, thus requiring the child to perform a mental rotation in order to imitate the correct placement of the object. Children aged 3-5 years generally have great difficulty in the rotation condition, and full success is not achieved until around 9-10 years of age. Of most interest to us, however, is the simplest condition in which the two spaces are identically aligned with each other. Are very young children able to copy in one space the placement of an object in a different space? To d o so, the child must recognize the correspondence between the spaces and must map the objects or places in one space onto those in the second. This task is obviously similar to the imitation trials in the familiarization phase of the model studies described previously. In the most extensive study of this sort, Laurendeau and Pinard (1970) presented children two identical miniature landscapes. Each contained a road and railroad tracks dividing it into four parts, as well as five toy houses of differing sizes and colors. On each of 12 trials, the experimenter placed a small clay man at one point in one of the landscapes and asked the child to place a comparable figure “at exactly the same place” in the second landscape. The points on the landscapes were chosen to test for the application

14

Judy S. DeLoache

of topological concepts, so they varied in terms of their proximity to landmarks. The youngest subjects in this research were 100 3- and 3.5-year-olds. A few (five) of the youngest children showed almost total incomprehension of the task (Stage 0). Some flatly refused to cooperate. Others are more patient and resign themselves, with no great enthusiasm, to going on with the test a little longer. Most of the solutions offered, however, unequivocally show that this apparent docility and good nature hardly make up for lack of ability or incomprehension. (p. 219)

Most of Laurendeau and Pinard’s (1970) 3- and 3.5-year-olds showed more comprehension of the task. The majority were classified in Stage 1-they “understand the nature of the task and apply themselves to it rather willingly” (p. 221) -but still without much actual success in placing the object correctly. One-third (32) of this age group reached Stage 2A: They were able to place the object correctly on the easier trials -that is, those with a salient topological cue. Stage 2B -success on positions that were not unambiguously specified by topological cues-was not reached until almost 2 years later (5.4 years). Although they could manage the unrotated landscapes, Stage 2 subjects were still unable to locate the object correctly when one space was rotated 180 degrees with respect to the other. Stage 3, characterized by success with the rotated boards, was reached around 10 years of age. How does the performance of Laurendeau and Pinard’s (1970) 3-year-olds compare to the behavior of the 2.5- and 3-year-old subjects in the model task? In one study (Debache, 1989), we standardized the familiarization phase of the task and recorded the accuracy of the children’s imitations of the experimenter’s placements of the toy (this phase was not systematically recorded in Study 1). We found that a group of 3-year-old children (36-39 months) successfully imitated on 67% of their familiarization trials. The 3-year-olds in Laurendeau and Pinard’s (1970) study seem to have achieved a similar level. According to my own calculations, based on their data (Table 30, p. 186), the 3-year-olds placed the figure correctly 37% of the time. If one includes only the easiest trials in which the figure was to be placed in close proximity to a salient landmark, the level of correct placement was 56070, reasonably comparable to our figure of 67% correct imitations. The slightly higher performance in the model studies is probably due in part to the realism of the spaces and the salience of the landmarks. Pufall and Shaw (1973) found that even children as old as 4.5 years had difficulty imitating object placements in nonrealistic spaces (e.g., two flat boards divided by lines into quadrants, each of which contained a geometric shape in the center and four pegs in the corners). The children placed the object in the wrong quadrant on approximately 50% of the trials even when the boards were aligned identically.

Represenrarion

15

Those few of Laurendeau and Pinard’s (1970) 3-year-old subjects who show no comprehension of the task are like our 2.5-year-old subjects in the model task. In both cases, the children seem unaware of the correspondence and systematic relationship between the two spaces. Spatial cognition studies thus agree with the model task in showing that mapping the places within one space onto the places in a second space can be problematic for very young children. The crucial factor seems to be the basic realization that the two spaces are related in a meaningful way. Without that fundamental realization, it is impossible to comprehend what the task is all about, let alone imitate the experimenter’s placement of a visible object or retrieve an invisible one. Once the child does recognize the correspondence between two spaces, his or her performance will vary as a function of how difficult it is to carry out the requisite mappings from one space to the other. In the model task, the individual places are highly salient objects and the relation between object and hiding place is simple and unambiguous. Thus, the whole problem lies in being aware of the correspondence between the spaces, and as a result, performance is all or none. In more complex situations, such as the perspective-taking tasks in which the stimuli are less realistic, the spatial relations are ambiguous, and the spaces are not aligned with one another, additional cognitive skills are required, and gradual developmental progress is observed.

VI. Analogical Reasoning in Young Children The foregoing has stressed that the challenge presented to young children by the model task is conceptual-realizing (1) that one space is in important ways like a second space that differs from it in numerous respects and (2) that what is known about the first space can (and should) be applied to the second. This is essentially an analogical reasoning problem. Considering the model task in this context may help to specify more precisely the relevant underlying processes. In analogical reasoning, knowledge of one domain-the base-is used to think about or solve a problem in a second domain-the target. More is known about the base than about the target, and one’s knowledge of what is true about the base is used to draw inferences about what might or must be true about the lesser-known target. Analogical transfer is possible when the knowledge representations of two different domains share certain features in common (cf. Gentner, 1980, 1983; Holyoak, 1984). According to Gentner’s (1983) formulation, the first step in analogical reasoning is gaining access to the analogy, that is, noticing that a correspondence exists between base and target. Accessibility or noticing is not a trivial problem; indeed, people of all ages often fail to solve problems by

16

Judy S. DeLoache

analogy simply because they do not recognize the relevance of a potential base domain to the target problem. Once the noticing step has occurred (once access has been gained to the analogy), correspondences are set up between the objects in the base and target. In the critical final step, relations among the objects are mapped from the base to the target. It is the mapping of relations that permits one to draw inferences about the target domain. With respect to the model task, the space in which the child observes the toy being hidden is the base and the second space is the target: The child knows more about the base (Le., he or she knows where the object is hidden) than about the target (the location of the invisible object in the target space must be inferred). (For ease of discussion, I continue the room-to-model example used before, referring to the room as the base and the model as the target, but the model was actually the base for half the subjects.) The older children in the model task succeed in mapping relations from the base to the target. They transfer the hiding relation-the relation between the hidden object and its hiding place-from the original space to the analogous space. Thus, they use their representation of where the original object was hidden to infer where the analogous object must be hidden. To do so, they must at a minimum have noticed the correspondence between the hiding places within the two spaces; one cannot map a relation among objects without detecting the correspondence between the objects. In addition, we know from their spontaneous comments (“They’re the same” or “just alike”) that these children also noticed the overall correspondence between the room and model. The younger subjects in the model task apparently performed none of the steps involved in analogical reasoning. Not only did they fail to map the hiding relation (which their Retrieval 2 performance shows they had represented) across domains, but there is no evidence that they ever established the object correspondences that would enable such transfer. Furthermore, the younger subjects gave no sign of understanding the basic correspondence between the overall spaces-that is, between the room and the model. In the terms of analogical reasoning, then, the fundamental problem of the younger children in the model task is one of gaining access to the analogy, of noticing the relation between the base and target space. The fact that the 2.5-year-old children in the model studies have so much trouble noticing the correspondence between the two spaces does not set them apart from other subjects in analogical reasoning research. In one of the bestknown series of studies on analogical problem solving, that of Gick and Holyoak (1980, 1983), one of the major findings was the surprising extent to which adults often fail to notice the similarity and relevance of one problem to another. When subjects fail to notice spontaneously the existence of an analogy, hints about the relation between two analogues are often sufficient to induce

Representation

17

access. Simply being told that two problems are alike often has a dramatic effect on the likelihood that both adults and children will perceive and apply an analogy. Although spontaneous transfer was unimpressive in the studies by Gick and Holyoak (1980, 1983), they found that their adult subjects were very responsive to hints. In one study, the proportion of subjects giving a solution went from 7% before the hint to 75% after it. Clearly, not recognizing the analogy was the major impediment to success in these studies. Children can similarly be helped in problem-solving situations by hints to apply a solution previously learned to a different problem. In a series of studies by Crisafi and Brown (1986), 2- to 4-year-old children learned a series of three structurally similar problems. In each problem, the child had to learn, first, how to obtain a token of some sort and, second, how to insert the token into an apparatus to make it dispense a reward. The problems varied in the familiarity both of the objects and of the relations among them (for example, they ranged from learning to get a coin from a purse to learning how to obtain a ball bearing from an automated box apparatus). Children who were informed and periodically reminded by the experimenter that, “All my games are candy games and you play them all the same way,” performed significantly better on the harder problems than did children not given the hints. Holyoak, Junn, and Billman (1984) also found an improvement in children’s problem-solving performance when they were given the hint that a story they had just heard could help them solve a problem. A developmental difference occurred in their subjects’ responsiveness to the hint, however. Once the analogy was pointed out, the ll-year-old subjects were almost always able to apply it. In contrast, the performance of a group of preschool children was highly variable. Sometimes a hint helped them succeed, but in a substantial number of cases, the hint was rejected, often because the children placed too literal an interpretation on it. For example, one young child interpreted very concretely (and inappropriately) the experimenter’s suggestion that the previous story could help solve the current problem of moving some balls from one location to another: The child “responded to the suggestion, ‘Could the story help?’ by retrieving two pictures that had been used to illustrate the story and using them to push the balls around” (p. 2051). This failure to profit from hints is similar to our findings. Our 2.5-year-old subjects seemed impervious to our efforts to establish the correspondence between the room and the model for them. Not only were these children unaware of the relation between the two spaces, but also, as far as we can tell, they were ineducable. We did not restrict ourselves to gentle hints about the relation between the two spaces; we fully and explicitly described it. However, the children failed to perform the mapping between corresponding objects, even when each miniature object was held up against the full-sized object while the experimenter described how they were the same. Thus, the question we posed earlier-What is responsible for the younger children’s

18

Judy S, DeLoache

failure in the model task?- could have been stated even more strongly: What prevents young children from grasping the correspondence between the room and the model, even after having the relation explicitly described and demonstrated? One cause of resistance to an analogical relationship is a focus on those aspects of the two analogues that are dissimilar, rather than on those that are the same. According to Gentner’s (1983) structure mapping theory of analogy, relations have precedence over object attributes in the process of formulating and comprehending analogies. Indeed, what characterizes nonliteral comparisons such as analogies is that identical relationships hold among nonidentical objects. Thus, in using an analogy, one must ignore the specific characteristics or attributes of the objects involved and focus instead on the relations that hold between or among the objects. This preference for relational over attribute matches emerges with the development of knowledge in a domain, as well as with general cognitive development. The generally impoverished state of knowledge of the young child - the “universal novice’’ (Brown & DeLoache, 1978) -often precludes their perceiving the deeper and less obvious structural relations among objects. Unlike older individuals, young children prefer and show better comprehension of metaphors based on attribute similarities than metaphors based on relational similarity (Gentner, 1986). When hints are effective in analogical reasoning situations, they presumably induce the subject to ignore surface dissimilarities in favor of underlying structural similarity. When told that two stories or games are alike, the subject may first try to match them on surface features but, failing that, then go on to look for matching relational structure. This analysis leads to two clear predictions, both of which have been confirmed in developmental studies of analogical reasoning. One is that any manipulation that highlights the structural similarity between two analogues should enhance transfer, especially for younger children or novices who are less likely to notice underlying relational similarities on their own. The second is that increasing the surface similarity between two analogues should facilitate analogical reasoning, again especially for those subjects unlikely to have spontaneous awareness of underlying similarity. Making the object correspondences more transparent should make it easier to map one onto the other, which should in turn facilitate mapping the corresponding relations among objects. Highlighting structural similarity has been shown to improve transfer in both adults and children. Gick and Holyoak (1983) enhanced transfer in adults simply by providing multiple-base analogues. They argued that this manipulation led the subjects to form an abstract schema of the structural elements common across the problems. Using a more extensive intervention, Brown, Kane, and Echols (1986) facilitated transfer in 3- to 5-year-old children by drawing their attention to

Representation

19

the structural relations common to a set of three stories. After each story, the child was asked to identify the protagonist of the story, the protagonist’s goal, the obstacle preventing attainment of the goal, and the solution. As expected, directing attention to the common goal structure of the stories facilitated transfer. Especially interesting results came from a recall condition, in which subjects were simply asked to tell everything they could remember about the story. Some of the children in this group spontaneously recalled all four of the goal structure elements, and most of them also succeeded on the transfer test. Few of the subjects who initially failed to recall the goal structure elements achieved the correct solution on the target problem. Their difficulty was not a simple memory deficit, however. When prompted for recall of the four structural elements, these subjects readily recalled them and, most important, this prompting then led to successful transfer by the majority of the children. Thus, processing the crucial structural features that the stories had in common, whether done spontaneously or in response to prompts, resulted in transfer. The second prediction mentioned above previously-that increased perceptual similarity should facilitate transfer-has also received empirical support in developmental research. Gentner and Toupin (1986) investigated the effect of physical similarity on children’s ability to establish the object correspondences between base and target stories. n o age groups of children (4- to 6-year-olds and 8- to 10-year-olds) listened to a brief story with animal characters in the roles of hero, friend, and villain, and then acted out the story using a set of toy animals. Next, the child was required to transfer the story plot from one set of characters to another. Object similarity was manipulated to make the correspondence between base and target objects easy or hard to see. The characters in the second set were very similar in appearance to the first set (chipmunk-squirrel, robin-bluebird, horse-zebra) or quite different in appearance (chipmunk-elephant, robin-shark, horse-cricket). In the hardest condition, crossed mapping, characters that looked similar to the original characters played different roles in the second story (chipmunk-zebra, robin-squirrel, horse-bluebird). Object similarity had the expected effect on the children’s mapping of the characters from the base to the target story. Having highly similar characters in the same roles resulted in around 90% correct enactments and essentially equal performance for the two age groups. With the less similar characters, and especially with the reversed characters, the younger subjects were considerably less accurate than the older ones. The performance of the older subjects also dropped off in those conditions, but less so. Thus, object similarity facilitated transfer for both ages; but, as predicted, it was more important for the younger children, presumably because they relied more on object attributes than did the older children. Brown and Kane (1986) also reported age differences in the effect of

20

Judy S. DeLoache

perceptual similarity on transfer. A younger group of kindergarten and secondgrade children and an older group of fourth-grade children heard three superficially dissimilar but structurally similar stories. The solution to all three stories was identical: A large piece of paper was rolled into a tube and used to transport some items across a barrier. In a high-physical-similarity condition, the paper was already rolled into a tube, so that it was physically identical across problems and its conveyance function was emphasized. In the low-similarity condition, the paper lay flat on the table. The two age groups showed the same excellent level of transfer in the highsimilarity condition. In the low-similarity condition, the older children were still very successful, but the younger ones were not. Thus, perceptual similarity had a clear effect on transfer, and the effect was especially pronounced for the younger subjects. The younger children needed the support of the identical tool to transfer from one story to another; older children were capable of arriving at the solution without such salient perceptual support. A high degree of surface similarity presumably aids both in gaining access to an analogy (that is, realizing that two domains are related) and in establishing object correspondences in the two domains once the overall relation has been noticed (Gentner & Toupin, 1986). Given the theoretical arguments and empirical evidence for the importance of perceptual similarity for analogical reasoning (Gentner & Landers, 1985; Holyoak & Koh, cited in Holyoak, 1984), especially for children (Brown & Kane, 1986; Gentner & Toupin, 1986; Holyoak et al., 1984), one would expect that perceptual similarity should be an important variable in a situation like the model task in which young children must perceive and use the correspondence between two spaces. Accordingly, two studies addressed the role of perceptual similarity in young children’s space-to-space reasoning. A. STUDY 3: SIMILARITY OF OVERALL SPACE VERSUS INDIVIDUAL PLACES

1. Procedure

In Study 3, we examined the independent contributions of similarity of the overall spaces and similarity of the hiding places within the spaces. In the original model studies, we varied the perceptual similarity of the individual hiding places, but performance was identical for the two highly similar and the two dissimilar locations. This result is somewhat puzzling, but it may be that the presence of some high-similarity hiding places was adequate to support transfer and that a more comprehensive manipulation of surface similarity would have the hypothesized effect. In Study 3, all the hiding places in the model were either highly similar or highly dissimilar to the hiding places in the room. In addition, the similarity of the surrounding spaces -the room

Representation

21

and the model-was varied to see if it affected young children’s space-tospace reasoning. The nature of the surrounding space can be very important in spatial cognition, both for animals (Gallistell, 1989) and for children (Keating, McKenzie, & Day, 1986). For these purposes, an artificial room was constructed. It measured 2.57 m x 1.85 m x 1.88 m high and was large enough for both a child and an adult to move around in. A framework of plastic pipes supported walls of opaque white fabric. An opening in the center of one wall served as the entry. Two models were constructed to correspond to the room, but to vary in their perceptual similarity to it. The high-similarity-of-space model (62.9 cm x 48.3 cm x 38.1 cm) was made of the same materials as the room-plastic pipes and white fabric. The low-similarity-of-space model (69.9 cm x 45.7 cm x 38.1 cm) was a cardboard box covered with white paper. Both models were open on the side on which the entry was located in the artificial room. The room was furnished with a dresser and a set of shelves made of heavy cardboard, a floor pillow, a basket, a chair covered with fabric, chair pillow, rug, and window with curtains. In the high-similarity-of-places condition, the model furniture was as similar as we could make it to the room furniture; and in the low-similarity-of-places condition, the items of furniture in the model were of the same categories as those in the room, but they were dissimilar in appearance. The subjects for Study 3 were 70 children (38 males and 32 females). Two age groups of 35 children each were comparable to the groups in Study 1: The younger group was 31 months of age, and the older group was 38. At each age level, there were four independent groups, resulting from the four combinations of high and low similarity of space and places. The procedure followed was basically the same as for the original Study 1. Each child watched as a toy was hidden in either the artificial room or one of the models, and then the child searched for the corresponding toy in the corresponding space.

2. Results Figure 4 shows the results of Study 3 and indicates that surface similarity did affect performance in the model task. The most important results were significant interactions of Condition x Age and Condition x Retrieval. (Figure 4 shows the data as a function of age, condition, and retrieval, even though the three-way-interaction was not significant, in order to facilitate comparison of these results with those of Study 1.) As expected, there were no differences anywhere for Retrieval 2. However, as is clear from the figure, Retrieval 1 performance was best when both the overall spaces and the individual hiding places within the spaces were highly similar and worst when they both were dissimilar. The similarity of the hiding places was clearly more

Judy S Debache

22

Old

20 0

t Retriml 1

Retrieval 2

Fig, 4. Brcentage errorless retrievals 0s afunction of age and similarity between the individual hiding places and the surrounding spaces of the model and room (Study 3).

important in supporting transfer than the similarity of the surround. The highplace-low-space condtion did not differ significantly from the high-high condition, but they were both significantly better than the conditions with low place similarity (which did not differ from each other). The effect of hiding place versus contextual space similarity was not equal for the two age groups, however. In the two conditions with high place similarity, the older subjects performed significantly better than the younger ones; but in the two low-place-similarity conditions, the two ages did not differ. This result indicates that perceptual similarity is quite important in the model task. If the actual hiding places do not look alike, transfer is virtually nonexistent. If the hiding places do resemble each other-if the chair in the model looks like the chair in the room-the older, but not the younger, children use this physical resemblance to establish the overall correspondence between the two spaces. Thus, access to the analogy, the correspondence between the room and model, is facilitated by surface similarity. The most dramatic aspect of the results of this study is the very poor Retrieval 1 performance of the older children in the low-place-similarity condition. Thus, the seemingly robust success of the older children in the standard model task (in Study 1 and subsequent studies) is actually fairly fragile. Their understanding of the correspondence between room and model can be almost totally disrupted by the absence of obvious, physical similarity between

Representation

23

the places in which the toys are hidden in the two spaces. The conceptual equivalence of the two spaces is, for 3-year-old children, still rooted in perceptual similarity. B. STUDY 4: IDENTICAL SPACES

In this study, we examined further the effect of similarity by increasing similarity to the maximum. We thought that asking young children to reason from one space to a virtually identical space should help them recognize the correspondence between the two spaces. This situation would represent a case of extremely high similarity, both of overall spaces and of hiding places. Furthermore, there would be no difference in scale, which has been suggested as a possible impediment to young children’s space-to-space reasoning (Acredolo, 1977; Siege1 et al., 1979). Based on the results of Study 3, we expected that young children’s performance would be facilitated with identical spaces. 1. Procedure In one condition, transfer was assessed from one full-sized room to another nearly identical one of the same size (4.80 m x 2.74 m x 2.45 m). They contained duplicate furnishings, including built-in bookshelves along one wall. The only difference between the rooms was that one of them had a one-way observation mirror behind the shelves, whereas the other had a mirror (curtained) on the opposite wall. For the second condition, children were required to transfer from one commercial dollhouse (54.61 cm x 25.40 cm x 30.48 cm) to another, identical dollhouse. In both conditions, the child watched as a toy was hidden in one space. Then the child walked down the hall to the other room (in the two-room condition) or around a partition to other other dollhouse. There were 32 subjects-8 girls and 8 boys in each condition. They were equivalent in age to the younger subjects in the original model study: All were between 29 and 31 months of age, and their mean age was 30 months. The performance of the children in the two identical-spaces conditions was compared to that of the younger group from Study 1.

2. Results The data are shown in Fig. 5, superimposed, for purposes of comparison, on the data from Study 1. The Retrieval 1 performance of the 2.5-year-old children in both of the identical-spaces conditions was significantly better than that of the comparable age group tested with spaces of different scale. Thus, Studies 3 and 4 both show that increasing the physical similarity between the base and target spaces facilitates transfer between them.

24

Judy S. D e b a c h e

Retrieval 1

Retrieval 2

IAnalogous Location1

IOriginal Location1

Fig, 5. Percentage errorless retrievals when two spaces are identical to each other (Study 4). The data from Siudy 4 are superimposed on those from Study I to facilitate comparison.

There were some surprising aspects to the data, however. The Retrieval 2 performance of the identical-spaces groups was significantly poorer than that of the children tested in the standard model task. This study is the only one of all the studies we have conducted with the model task in which performance on the memory retrieval has been lower than the expected level of around 80%. The Retrieval 1 data for Study 4 are also somewhat puzzling when compared to the results of Study 3. In both experiments, the 2.5-year-olds did better than their age-mates in the standard model task (41 and 43% versus 14%, respectively), thus showing a positive effect of increased physical similarity. However, if perceptual similarity increases performance in the model task, as it apparently does, why was identicality not more advantageous than high similarity? Why were the identical-spaces groups of Study 4 not higher than the high-high similarity group of Study 3? The surprising aspects of the children’s performance in Study 4 suggest that although increasing perceptual similarity had positive effects on performance, identicality may have had some negative impact. The lower than expected scores, especially for Retrieval 2, may in part reflect an interference effect, the well-known phenomenon of increased interference with increased interlist similarity. I suspect that the extremely high degree of similarity was actually somewhat disorienting, sometimes causing the children to lose track of the current hiding place. Another aspect of the identical-spaces conditions that may have affected the children’s performance is the fact that it is a rather anomolous situation. It is not a proper analogy, in that the base and target are identical. This may have struck even our 2.5-year-old subjects as somewhat peculiar, and, in fact, a number of children showed quite odd search patterns in this study, especially in the two-rooms condition. Several subjects perseverated on a single location, repeatedly searching there before looking anywhere else. On a few occasions,

Representation

25

the children appeared to know full well the toy was not in a given place, but they searched there anyway. For example, a 30-month-old girl said, “There’s no one in there,” before looking in the box that she had searched first on most of the preceding trials. This kind of stereotyped searching may represent a way of dealing with a situation that the child does not understand. These two studies thus reveal that increasing perceptual similarity has a positive effect-up to a point. Just as is unfortunately so true in many realms of life, one can get too much of a good thing. There must be some difference between two analogues or they are not analogues at all. Differences in scale between the two spaces mark the model task as an analogical reasoning problem. Thus, rather than having a deleterious effect on performance, as was hypothesized based on the results of Siege1 et al. (1979), differences in scale may actually play a positive role when children are asked to reason from one space to another. One important question continues to loom large: What is responsible for the younger children’s failure to comprehend the relation between the two spaces? To try to answer this question, 1 now turn to the second sense of representation described in the beginning of this article-representation in the narrow sense, or symbolic representation. In the following section, young children’s performance in the model task is examined with respect to the development of symbolization. Both senses of representation turn out to be operative in young children’s space-to-space reasoning.

VII.

Symbolization in Young Children

The model task is a symbolization problem. What it requires is that the child understand that the model represents or stands for the room. In other words, the model serves as a symbol for the room: “Symbolization is the representing of an object or event by something other than itself’ (Potter, 1979, p. 41). Scale models-like drawings, photographs, maps, sculptures-are symbolic representations. “A symbol brings to mind something other than itself’ (Huttenlocher & Higgins, 1978, p. 109). The younger children’s difficulty with understanding that the model represents the room cannot be due to a general absence of symbolic competence. These children have been using language for some time. They have presumably engaged in symbolic play, including object substitutions (Ungerer, Zelazo, Kearsley, & O’Leary, 1981). In addition, children of this age have much experience with various symbolic media, including books and television. From infancy onward, children perceive the correspondence between real objects and photographs or drawings of them (Debache, Strauss, & Maynard, 1979; Dirks & Gibson, 1977; Hochberg & Brooks, 1962). Although infants do have

26

Judy S. Debache

to learn about the special status of books (i.e., that they are not to manipulate or eat-Murphy, 1978), even very young children understand a great deal about the representational nature of books and pictures. A. YOUNG CHILDREN’S UNDERSTANDING OF MAPS

Maps are, of course, a good example of a symbolic medium; and the small but growing literature on the development of symbolic spatial cognition, including the understanding and use of maps (e.g., Presson, 1982), is of particular relevance to this article. Using a map to locate a position or object in a real space is similar to what is involved in the model task; a map represents a space just as a model does, and the features on the map represent places within the space. In a study that is conceptually very close to the model task, Bluestein and Acredolo (1979) investigated young children’s map use. Three- to 5-year-old children were shown a simple map of a room. Each feature of the room was represented by a colored line drawing of the object in its proper position. A small toy was placed on the map, and the child was instructed to use the “picture” (the map) to find a larger toy in the room. The 3-yearold subjects were only marginally competent map readers. Only half of them managed to find the toy in the easiest condition-with the map inside the room and perfectly aligned with it. Thus, a relatively abstract symbolic representation of a space, such as a map, is apparently even more difficult for young children than is a model. Downs and Liben (1986) have been studying many aspects of the development of children’s understanding and use of maps. Their preliminary findings indicate that although preschool children have definite concepts about what does and does not constitute a map, their ability to interpret and use maps is at best primitive. They provide several charming and thought-provoking examples of confusion on the part of young children who were attempting to interpret a road map of Pennsylvania: While some children correctly identified Lake Erie as water “because water is blue,” this same reasoning led some children to think that if they stood on a road shown in red on the map, the road itself would actually look red. Similarly, the airplane symbol was interpreted as showing an airplane. If that particular airplane were to fly away, there would be no plane on the map. (p. 21)

The main problem revealed in these examples is that the young children are treating the symbols on the map not as pure symbols, but as iconic representations. Lakes are blue, so a large blue area must be a body of water. An outline of an airplane must represent a real airplane. My guess would be that these young children are trying to apply the rules and procedures they know for interpreting pictures and photographs to the road map. As a result, they

Representation

21

expect much more concordance between the representation (the map) and its referent than actually exists. Downs and Liben (1986) reported other kinds of interesting errors made by young children when trying to interpret maps. Responses to an aerial photograph of Chicago were somewhat piecemeal; individual objects were identified, but the children’s perception of the represented objects was not constrained by the relations that exist among those objects in the real world. My favorite example came from a girl who, “having moments earlier identified a road correctly, pointed to a semicircular shape near it (a grassy area) and identified it as ‘cheese”’ (p. 19). In Gentner and Toupin’s (1986) terms, this young child, in mapping the aerial photo onto a real-world space, focused on individual object attributes (the semicircle looked like a half-round of cheese) rather than on relations among objects. However, the relations among objects (in this case, size relations) should constrain the possible interpretations: If one feature on the map is a road, then an area that is relatively large with respect to it can almost certainly not be a hunk of cheese (on the moon perhaps, but not on earth). Young children thus show quite limited skill in dealing with abstract spatial representations. Three-year-old children, who have mastered the model task, still show little competence with maps. The concrete, three-dimensional nature of a model apparently makes it more comprehensible to 3-year-olds than a map, which has little or no physical resemblance to its referent. The precise nature of the symbolic representation of a space will have a large effect on young children’s ability to interpret and apply it. B. MODEL AS SYMBOL: THE DUAL REPRESENTATION HYPOTHESIS

Let us now return to the model task and ask again what might be the source of the 2.5-year-old children’s extreme difficulty with it. The conclusion that I have reached may at first seem contradictory to the preceding argument, but in the end, it is not. I wish to suggest that the poor performance of the younger children in the model task stems from a peculiar limitation on their symbolic capabilities - in particular, their ability to treat a real object as a symbol. To succeed in this task, the child must have a dual orientation to the model. For one thing, the model is a real, three-dimensional object-actually, a set of objects with which the child interacts. At the same time, the child must see the model as a representation of something else, as a symbol for the room. I hypothesize that the difference between the performance of the older and younger subjects in the model studies arises from the need to represent or respond to the model in two different ways at the same time. The younger children may see the model only as a real thing and not as a symbol of

28

Judy S. &Loache

something other than itself. Real objects, especially objects with functions of their own, typically d o not serve as symbols for other objects. As Potter (1979) said, “HOWis it that symbols such as words and pictures represent real objects, whereas real objects don’t represent anything-they simply are objects?” (p. 57). In addition to being unfamiliar with objects used as symbols, young children may not possess the cognitive flexibility to think simultaneously of the model both as a real thing in itself and as a representation of something else. If this analysis is correct, then 2.5-year-old children should do better on a version of this task that involves a different, strictly representational medium. C. STUDY 5: YOUNG CHILDREN’S USE OF PHOTOGRAPHS VERSUS MODEL

To test this hypothesis, the model task was modified so that photographs, rather than the model, were used to give the child the information about the location of the hidden object in the room. A two-dimensional photograph or other picture is a representation of something else, and it is only a representation. Its primary role is not that of a real thing, a thing that is manipulated or used in some way other than as a representation. To the extent that young children have trouble using objects as symbols, their retrieval of the hidden toy based on the photographs should be better than their model-based retrieval. One appealing aspect of this prediction is that it is counterintuitive on other grounds. Generally, two-dimensional stimuli are thought of as impoverished relative to three-dimensional stimuli, which are considered to be richer, more salient, more informative. Infants and young children typically learn better with real objects than with pictures (e.g., Daehler, Lonardo, & Bukatko, 1979; DeLoache, 1986; Hartley, 1976). Cross-cultural studies also have shown advantages for objects over pictures in a variety of cognitive tasks. For example, Scottish and Zambian school children performed equally well when asked to sort real objects into categories, but the Zambian children fared less well when required to classify photographs of the objects (Deregowski & Serpell, reported in Cole & Scribner, 1974). 1. Procedure In Study 5 , sixteen 30- to 33-month-old children (8 males and 8 females) served as subjects. Their mean age of 31.6 months was comparable to that of the younger group in Study 1. The base information for this task consisted of four color photographs of the same laboratory playroom (Fig. 1) used in Studies 1 and 2. Each photograph showed one or two of the hiding places. The four photographs

Representation

29

were arranged in a semicircle on top of a table in the small room adjoining the playroom. At the beginning of the session, the Snoopy toy was introduced, and the correspondences between the pictures and the furniture in the room were pointed out. The child was told that Snoopy wanted to play a hide-andseek game with him or her. For each of the four trials, the child waited in the small adjoining room while the toy was being hidden in the playroom. Then the experimenter pointed to one of the four photographs. She pointed directly at the hiding place, and said, “He’s hiding back [under] here.’’ Then the child was asked to find the toy in the room. In order to have a direct comparison of performance in the photograph and model tasks, each child was tested on two different days. Half the children received the photograph task on the first day and the hide-in-model-retrievein-room task on the second day. The other half experienced the two tasks in the opposite order. 2.

Results

The results for Study 5 are shown in Fig. 6. The point of most interest is the children’s performance in the photograph condition; they achieved 70% errorless retrievals. This is the highest score for any group of younger subjects in the whole series of studies reported so far. It is fully comparable to the Retrieval 1 performance of the older children in the original model study. The model task data for these same children are also shown in Fig. 6. The most important thing to note about the data from the model condition is that the children’s Retrieval 1 performance was substantially and significantly below their performance in the photograph condition. The same child who

- 100v)

Ba,

c L

80-

QI

cL

60-

v) v)

QI

z 2

40-

kl 8

20A Wide-Angle

0

Retrieval 1

Retrieval 2

IAnalogous Location)

10riginal Location)

Fig. 6. Percentage errorless retrievals using photographs versus a model (Study 5).

30

Judy S DeLoache

was unable to find the toy after seeing it hidden in the model was able to find the toy after seeing a photograph of its hiding place. There was no effect of order of task. (Note that the performance of the 2.5-year-old children in Study 5 almost exactly replicated the performance of the same age group in Study 1.) One potential criticism of this study is that a photograph of an individual object is such a simple and iconic representation that it may simply designate or provide direct access to the relevant hiding place, rather than serve as a symbol of it. In other words, showing individual photographs of single hiding places may be too much like simply pointing to the relevant location to provide an adequate test of our hypothesis. Accordingly, a second photograph study was conducted using a wide-angle photograph rather than the set of photographs of individual hiding places. The wide-angle photo showed approximately two-thirds of the room, including all the hiding places. The relative spatial positions of the objects were preserved, but the spatial relations were slightly distorted by the wide-angle lens. The design of this second experiment was the same as the preceding photograph study. The results were also the same (see Fig. 6). Errorless retrievals were 81% with the wide-angle photograph, slightly higher even than the 70% with the single photographs. This level of performance was significantly better than what the same subjects achieved in the standard model task. Thus, the superior performance with a two-dimensional, pictorial representation holds not just for photographs that depict individual hiding places, but also for a single photograph representing the entire space. These photograph studies show that in the original model studies, it was apparently the three-dimensional nature of the model that blocked the younger children from perceiving its symbolic role. They were unable to inhibit their dominant response to the model as a real, manipulable thing in order to see it also as a representation of something else. This interpretation is consistent with Potter’s (1979) description of the development of symbol interpretation: The ability t o interpret all three facets of meaning of a symbol-its conceptual referent, its own qualities as an object, and the relation between the two-is initially absent in a child. . . . A child’s limited ability to hold two or more things in mind simultaneously may prevent him from realizing the relation between the properties of the symbol and those of the referent. Either he focusses on the conceptual referent, or he looks at the symbol as an object: The sculpture is a big stone. [Development occurs in] the ability to apprehend the identity of the symbolic medium at the same time as the identity of the referent. (p. 60)

D. SYMBOLIC PLAY

The preceding analysis suggests that a major change in cognitive flexibility must be responsible for the abrupt shift in performance in the model task between 2.5 and 3 years of age. If so, one would expect to find related

Representation

31

developments during this period. One particularly good candidate is symbolic play, especially object substitution. When a young child uses a stick as a horse, the child’s action is inspired and regulated by his or her own ideas more than by the properties of the stick itself (Vygotsky, 1978). This action is symbolic, because the child is treating the stick as something other than what it is. The child’s normal behavior toward the object is suppressed in favor of the behavior appropriate to the referent object. Some theorists have characterized this behavior as requiring the coordination of multiple representations (e.g., McCune-Nicholich, 1981). According to Leslie (in press), underlying object substitutions and some other forms of pretense are a primary representation of the object-a veridical representation based on current perception-and a metarepresentation-a representation of the object decoupled from its normal meaning. Object substitution in pretend play and success in the model task both seem to require what I have been referring to as a dual orientation to an object. Hence, it should not be surprising to find that object substitution undergoes substantial development in the age range during which the model task is mastered-roughly between 32 and 37 months. The particular aspects of object substitution that develop during this period fit well with the argument being made here. For one thing, although young children are capable of some forms of object substitution prior to this age, they become increasingly less dependent on perceptual support in substituting one object for another. For example, 2.5-year-olds’ substitutions tend to be restricted to objects that bear some physical resemblance to the represented object (e.g., using a rectangular block of wood like a telephone receiver). Three-year-olds are able to perform more advanced object substitutions, ones with less physical support for the child’s action (e.g., making combing motions with a baby bottle) (Elder & Pederson, 1978; Ungerer et al., 1981). A second, related developmental trend during this period-one that is particularly important in the present context-is that young children become increasingly able to use a familiar object with known functions as a substitute for another object. Initially, only things with ambiguous functions (e.g., blocks) are used as substitute objects. A 2.5-year-old is likely to refuse a request to comb his hair with a tennis ball, saying, “I can’t. It’s a ball.” (Elder & Pederson, 1978). Children gradually become capable of ignoring what they know about something (e.g., that it is used in a particular way) in order to treat it as something else (e.g., use it in a different way). Pederson, RookGreen, and Elder (1981) asked 2.5- and 3-year-old children to imitate an experimenter’s performance of a familiar action using substitute objects that varied in terms of how likely they were to elicit a specific response. For example, the child might be asked to pretend to drink from a dustpan (a low-response object) or a pair of scissors (a high-response object). Both age groups readily pretended with the low-response objects. However, only the 3-year-olds were

32

Judy S. DeLaache

able to inhibit their spontaneous response to the high-response objects in order to use them in a novel way. The 2.5-year-old subjects had “difficulty transforming one object into another if the substitute object [had] a high likelihood of eliciting manipulative responses” (p. 759). The literature on symbolic play thus shows that between 2.5 and 3 years of age, young children become increasingly able to take an object that has a known function and use it as something else. At the same time, children become capable of responding to a scale model of a room, not just as a set of play objects, but also as a representation of something else. A related example of young children’s difficulty in seeing a single object in two different ways comes from Susan Carey (personal communication, 1985). Young children were observed playing in a dollhouse in which the furniture and dolls had been set up for a tea party. The experimenter announced to the child, “We need some cups for tea.” Then the experimenter gave the child a shot glass, asking, “Is this a big glass or a small glass?” If the child said “small,” the experimenter persisted, “Is it big or small for the doll?” Threeyear-old children might initially say either that the glass was big or that it was small. However, if they had answered that it was small, they immediately responded to the experimenter’s query by indicating that it was “big for the doll.” In contrast, children only a few months younger (33 months) did not amend their original judgment in response to the experimenter’s probe. If they originally said that the glass was small, they insisted that it was the same for the doll: “I said it was small.” The 3-year-old can apparently represent the glass in two ways, the younger child in only one. This review suggests that the reason the model task is so very difficult for 2.5-year-old children is that it requires that they represent the model in two different ways. At the same time that the experimenter and child are manipulating parts of the model, behaving toward it as a thing, the model must also serve as a symbolic representation of the room. It is specifically this dual role of the model that eludes the 2.5-year-old children. This child represents and understands his or her experience with the model itself, but that knowledge is kept compartmentalized. The child knows where the miniature toy is hidden in the model but not where the corresponding toy is hidden in the room, because the child’s representation of the model has nothing to do with his or her representation of the room. What seems to be crucial, then, is the ability to represent, to think about, one and the same thing in two different ways. This ability involves a great deal of representational flexibility, more flexibility than the 2.5-year-old child can readily muster.

VIII. Related Research on Multiple Representations The cognitive flexibility necessary to form and coordinate two internal representations of a single meaningful object appears to develop between 2.5

Representation

33

and 3 years of age (at least in the relatively restricted populations studied so far). Further refinement of this representational flexibility occurs in the following year, as evidenced by research on the development of the appearance-reality distinction and the understanding of false belief. Flavell and his colleagues (Flavell, 1986; Flavell, Flavell, & Green, 1983) have established that young children make frequent errors when forced to distinguish between the momentary appearance and the reality of an object. For example, consider a 3-year-old child who is shown an object that looks like a solid rock. Upon manually examining the object, the child discovers that it is actually a soft sponge. When asked, “What is this, really and truly,” the child is likely to answer correctly that it is a sponge. However, the child is likely to give that same response, this time incorrectly, to the question, “When you look at this, what does it look like to your eyes right now?” In other words, the child fails to draw a distinction between what the object appears to be and what the child knows the object to be. In this example, the young child is making what Flavell (1986) refers to as an “intellectual realism” error-the child’s knowledge of the reality of the object dominates his or her judgment of what it looks like. The opposite sort of error-“phenominism”-is also made, especially with certain types of stimuli. In this case, the child’s perception of the appearance of an object dominates his or her judgment about its permanent nature. Flavell (1986; Flavell et al., 1983) believes that young children’s surprising degree of difficulty with the appearance-reality distinction reflects a metacognitive limitation: We adults easily resolve the seeming contradiction by identifying one representation of [an object’s] property or identity with its present appearance and the other with its reality. We identify the one with what we see and the other with what we know. . . . Although aware that external objects themselves cannot simultaneously be two different things at once, we are also aware that we can represent them as simultaneously looking like the one thing (“that’s what it looks like”) and really being the other (“that’s what it really is”). . . . young children are less cognizant of these facts about subjectivity and mental representation than older children and adults are.. . . they may try only t o decide what single thing the object ‘‘is,’’ as an entity out there in the world. That the object can be represented as having more than one “is,” inside our heads, may be a possibility that does not, or perhaps even cannot, occur to them. (Flavell, 1986, p. 10)

A related domain is the development of false belief attribution (Hogrefe, Wimmer, & Perner, 1986; Wimmer & Perner, 1983), the development of the understanding that a person can have and act on a belief that is at variance with what the child knows to be reality. In one scenario (Perner, Leekam, & Wimmer, 1987), young children listened to a story as the experimenter simultaneously acted it out with a model and miniature objects. The boy in the story places a piece of candy in one location (drawer) and goes out to play. In his absence, his mother moves the candy to a different location

34

Judy S. Debache

(cupboard). The boy then returns, wanting his candy. The subject was asked to indicate where he or she thought the protagonist would look for the candy. A dramatic increase appeared between 3 and 4 years of age in the likelihood that children would correctly say that the protagonist would go to the (empty) drawer. Young 3-year-olds (3.0-3.5 years) responded correctly only 21% of the time. They wrongly predicted that the boy in the story would go to the place where they knew the candy actually was (cupboard), not taking into account the fact that the boy could not know the candy was no longer where he had placed it (drawer). In contrast to the very poor performance of the 3- to 3.5-year-olds, 4- to 4.5-year-old children performed extremely well - 87% correct responses. Children between 3.5 and 4 years gave 60% correct answers. Perner et al. (1987) interpret their younger subjects’ behavior as an inability to attribute a false belief to another person, and they argue that this phenomenon reflects a “basic conceptual limitation in 3-year-old children’s understanding of the mind” (p. 19). To answer correctly in their task, a child must realize that the protagonist’s representation of the location of the candy would differ from the child’s own representation. Generally, the child must understand that two people can have contradictory beliefs; specifically, the child must understand that a person can have a counterfactual belief. What develops, then, is children’s ability to represent not just the situation presented to them, but also to represent another person’s representation of that situation and to accept that that representation may conflict with their own. Perner et al. (1987) point out that their work on the understanding of false belief and Flavell’s work on the appearance-reality distinction both involve the ability to deal with representations with conflicting truth values. In the case of the appearance-reality distinction, the child must realize that one individual him- or herself can have two different and contradictory representations of a single reality. In the case of false belief, the child must realize that two different people can have conflicting representations of the same reality. As would be expected from the conceptual similarities between these two domains, development in them follows a similar timetable: Children become capable of appropriately distinguishing between appearance and reality at about the same age that they can correctly attribute a false belief to another person. Understanding the representational role of scale models is like the ability to make the appearance-reality distinction and the ability to attribute false belief in that all three involve the coordination and understanding of multiple representations of a single reality. Understanding a scale model is conceptually simpler and develops earlier, because there is no conflict or contradiction between the child’s necessary representations. The model differs from the room in numerous aspects, but they are not contradictory. The differences between them are continuous; the two spaces are more or less similar on various dimensions. The distinction between the appearance and reality

Representation

35

of an ambiguous object, however, is discontinuous. The ambiguous object cannot be located somewhere on a continuum ranging from rock to sponge. Similarly, there is no middle ground between the belief states of individuals with different access to the facts of a situation. One similarity in the behavior of children in these three domains that is particularly striking is the ineffectuality of various forms of instruction and experience. Children in all three tasks are very resistant to efforts to improve their performance. I have described how our attempts to clarify for our 2.5-year-old subjects the relation between the room and model came to naught. Flavell (1986) has similarly reported that 3-year-olds in appearance-reality tasks remain unmoved by explicit instructions and intensive training. The same imperviousness to experience was observed by Perner el al. (1987). In their first experiment, children were given two stories involving false belief. The younger (3-year-old) subjects failed to do any better on the second than on the first story. In contrast, most of the few 4-year-olds who were wrong the first time profited from their experience and responded correctly the second time In addition, another group of 3-year-olds was not assisted by a manipulation Perner et al. (1987) designed specifically to improve their performance. Even after first having direct experience of being misled by a false belief of their own, they failed to attribute false belief and its consequences to another child. Children of different ages thus exhibit similar resistance in tasks that require them to have and understand multiple representations of a single object or event.

IX. Summary and Conclusion The research summarized here shows that young children undergo an abrupt transition in their ability to understand the relation between a scale model and a larger space. Virtually none of our 2.5-year-old subjects seemed to understand the relationship between the model and the room; almost all of the 3-year-olds did understand it. The difference seems to be that the 2.5-yearolds do not respond to the model both as a real thing and as a representation of something else. Its status as a complex, meaningful real object prevents their apprehension of its abstract relation to the room. The resistance to instruction, abrupt developmental shift, and negligible individual differences in the model task suggest the possibility of a strong maturational underpinning (Espenschade & Eckert, 1967). Further research, including a longitudinal study and cross-cultural comparisons, will be addressed to the issue of the role of experience in the development of mastery of the model task. The primary contribution of this research lies in the revelation of a hitherto undocumented abrupt developmental shift in very young children’s

36

Judy S. DeLoache

representational flexibility- in their ability to form and coordinate multiple representations. The ability to think of one thing in two ways is an important aspect of early symbolic development. In Western cultures, where so much of a child’s learning occurs via various representational media, this is a crucial step. More generally, the cognitive advance that occurs between 2.5 and 3 years of age provides a foundation for further developments in the understanding of multiple representations. ACKNOWLEDGMENTS This research was partially supported by Grant HD-05951 from the National Institute for Child Health and Human Development. Revision of the manuscript was partially supported by a Senior International Fellowship from the Fogarty Foundation of the National Institute of Health. The revision was accomplished while I was a Visiting Scholar at the Department of Experimental Psychology at the University of Oxford, England. I wish to thank Renee Baillargeon and Dedre Gentner for their encouragement and their close reading of earlier drafts of this manuscript. I also thank Usha Goswami for her comments on the article. I am indebted to Kathy Anderson, Valerie Kolstad, Debbie Kresser, and Mary Wraight for testing the children in the research reported here.

REFERENCES Acredolo, L. P. (1977). Developmental changes in the ability to coordinate perspectives of a largescale space. Developmental Psychology, 13, 1-8. Blades, M., & Spencer, C. (1986). Map use by young children. Geography, 71, 47-52. Bluestein, N., & Acredolo, L. (1979). Developmental changes in map-reading skills. Child Development, 50, 691-697. Brown, A. L., & DeLoache, J. S. (1978). Skills, plans and self-regulation. In R. Siegler (Ed.), Children’s thinking: What develops? @p. 3-35). Hillsdale, NJ: Erlbaum. Brown, A. L., & Kane, M. J. (1986). Analogical transfer in children: Conditions that promote functional furedness or flexibility. Unpublished manuscript, University of Illinois, Urbana. Brown, A. L., Kane, M. J., & Echols, C. H. (1986). Young children’s mental models determine analogicaltransfer across problems with a common goal structure. CognitiveDevelopment, 1, 103-121. Chi, M. T.H. (1978). Knowledge structures and memory development. In R. S. Siegler (Ed.), Children’s thinking: What develops? (pp. 73-96). Hillsdale, NJ: Erlbaum. Cohen, S. R. (1985). The development of constraints on symbol-meaning structure in notation: Evidence from production, interpretation, and forced-choice judgments. Child Development, 56, 177-195. Cole, M., & Scribner, S. (1974). Culture and thought. New York: Wiley. Crisafi, M. A., & Brown, A. L. (1986). Analogical transfer in very young children: Combining two separately learned solutions to reach a goal. Child Development, 57, 953-968. Daehler, M. W., Lonardo, R., & Bukatko, D. (1979). Matching and equivalence judgments in very young children. Child Development, 50, 170-179. DeLoache, J. S. (1986). Memory in very young children: Exploitation of cues to the location of a hidden object. Cognitive Development, 1, 123-137.

Representation

37

DeLoache, J. S. (1987). Rapid change in the symbolic functioning of very young children. Science, 238, 1556-1557. DeLoache, J. S. (1989). Young children’s understanding of the correspondence between a scale model and a larger space. Cognifive Development, 4, 121-129. DeLoache, J. S., & Brown, A. L. (1979). Looking for Big Bird: Studies of memory in very young children. Quarterly Newsletter of the Laboratory of Comparative Human Cognition, 1,53-57. DeLoache, J. S., & Brown, A. L. (1983). Very young children’s memory for the location of objects in a large scale environment. Child Development, 54, 888-897. DeLoache, J. S., & Brown, A. L. (1984). Where do I go next? Intelligent searching by very young children. Developmental Psychology, 20, 37-44. DeLoache, J. S., Cassidy, D. J., & Brown, A. L. (1985). Precursors of mnemonic strategies in very young children. Child Developmenf, 56, 125-137. DeLoache, J. S., Strauss, M., & Maynard, J. (1979). Picture perception in infancy. Infanf Behavior and Development, 2, 17-89. Dirks, J., & Gibson, E. (1977). Infants’ perception of similarity between live people and their photographs. Child Development, 48, 124-130. Downs, R. M., & Liben, L. S. (1986). Children’s understanding of maps. In P. Ellen & C. ThinusBlanc (Eds.), Cognitive processes and spatial orientation in animal and man: Vol. I . Neurophysiology of spatial knowledge and developmental aspects. Dordrecht, Holland: Martinus Nijhoff. Elder, J. L., & Pederson, D. R. (1978). Preschool children’s use of objects in symbolic play. Child Development, 49, 500-504. Espenschade, A. S., & Eckert, H. M. (1967). Mofor devetopmenf. Columbus, OH: Merrill. Flavell, J. H. (1986). The development of children’s knowledge about the appearance-reality distinction. American Psychologist, 41 418-425. Flavell, J. H., Flavell, E. R., & Green, F. L. (1983). Development of the appearance-reality distinction. Cognitive Psychology, 15, 95-120. Freeman, N. H. (1980). Strafegiesof represenlafion in young children: Analysis of spatial skills and drawing processes. London: Academic Press. Gallistell, C. R. (1989). The organization of learning. Cambridge, MA: MIT Press. Gentner, D. (1978). What looks like a jiggy but acts like a zimbo? Papers and Reports on Child Language Development, 15, 1-6. Gentner, D. (1980). The sfructure of analogical models in science (BBN Rep. No. 4451). Cambridge, MA: Bolt, Beranek, & Newman. Gentner, D. (1983). Structure-mapping: A theoretical framework for analogy. Cognifive Science, 7, 155-170. Gentner, D. (1986). Structure-mapping in the development of mefaphor. Unpublished manuscript, University of Illinois, Urbana. Gentner, D. (1988). Metaphor as structure mapping: The relational shift. Child Development, 59, 47-59. Gentner, D., & Landers, R. (1985). Analogical reminding: A good match is hard to find. In Proceedings of the International Conference on Systems, Man, and Cybernetics (pp. 607-61 3). Tucson, AZ. Gentner, D., & Toupin, C. (1986). Systematicity and surface similarity in the development of analogy. Cognitive Science, 10, 277-300. Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive Psychology, 12, 306-35 5 . Gick, M. L., & Holyoak, K. J. (1983). Schema induction and analogical transfer. Cognitive Psychology, 15, 1-38. Hartley, D. G. (1976). The effects of perceptual salience o n reflective-impulsive performance differences. Developmental Psychology, 12, 218-225.

38

Judy S. DeLoache

Hochberg, J., & Brooks, V. (1962). Pictorial recognition as an unlearned ability: A study of one child’s performance. American Journal of Psychology, 75, 624-628. Hogrefe, G. J., Wimmer, H., & Perner, J. (1986). Ignorance versus false belief A developmental lag in attribution of epistemic states. Child Development, 57, 567-582. Holyoak, K. J. (1984). Analogical thinking and human intelligence. In R. J. Sternberg (Ed.), Advances in the psychology of human intelligence (Vol. 2, pp. 199-230). Hillsdale, NJ: Erlbaum. Holyoak, K. J., Junn, E. N., & Billman, D. 0. (1984). DeveIopment of analogical problem-solving skill. Child Development. 55, 2042-2055. Huttenlocher, J., & Higgins, E. T. (1978). Issues in the study of symbolic development. In W. A. Collins (Ed.), Minnesota Symposia on Child Psychology (Vol. 11, pp. 98-140). Hillsdale, NJ: Erlbaum. Karmiloff-Smith, A. (1979). Micro- and macro-developmental changes in language acquisition and other representational systems. Cognitive Science, 3, 91-118. Keating, M. B., McKenzie, B. E., & Day, R. H. (1986). Spatial localization in infancy: Position constancy in a square and circular room with and without a landmark. Child Development, 57, 115-124. Kosslyn, S. M. (1978). The representational-development hypothesis. In P. A. Ornstein (Ed.), Memory development in children. Hillsdale, NJ: Erlbaum. Laurendeau, M., & Pinard, A. (1970). The development of the concept of space in the child. New York: International Universities Press. Leslie, A. M. (1988). Pretense and representation in infancy: The origins of “theory of mind.” Psychological Review, 94, 412-426. Liben, L. S., Patterson, A. H., & Newcombe, N. (Eds.). (1980). Spatial representation and behavior across the lifepan: Theory and application. New York: Academic Press. Mandler, J. (1983). Representation. In F. J. Flavell & E. M. Markman (Eds.), Handbook of child psychology: Vol. 3. Cognitive development (pp. 420-494). New York: Wiley. Mandler, J. M., & Johnson, N. S. (1977). Remembrance of things parsed: Story structure and recall. Cognitive Psychology, 9, 111-151. McCune-Nicholich, L. (1981). Toward symbolic functioning: Structure of early pretend games and potential parallels with language. Child Development, 52, 785-797. Murphy, C. M. (1978). Pointing in the context of a shared activity. Child Development, 49,371-380. Nelson, K., & Gruendel, J. M. (1981). Generalized event representations: Basic building blocks of cognitive development. In M. E. Lamb &A. L. Brown (Eds.), Advances in developmental psychology (Vol. 1, pp. 131-158). Hillsdale, NJ: Erlbaum. Palmer, S. E. (1978). Fundamental aspects of cognitive representation. In E. Rosch & B. B. Lloyd (Eds.), Cognition and categorization (pp. 259-303). Hillsdale, NJ: Erlbaum. Pederson, D. R., Rook-Green, A., &Elder, J. L. (1981). DevelopmentalPsychology, 17, 756-759. Perner, J., Leekam, S. R., & Wimmer. H. (1987). Three-year-olds’ difficulty with false belief The case for a conceptual deficit. British Journal of Developmental Psychology, 5, 125-137. Piaget, J., & Inhelder, B. (1956). The child‘s conception ofspace. London: Routledge & Kegan Paul. Potter, M. C. (1979). Mundane symbolism: The relations among objects, names, and ideas. In N. R. Smith & M. B. Franklin (Eds.), Symbolic functioning in childhood (pp. 41-65). Hillsdale, NJ: Erlbaum. Presson, C. C. (1982). The development of map-reading skills. Child Development, 53, 196-199. Presson, C. C., & Somerville, S. C. (1985). Beyond egocentrism: A new look at the beginnings of spatial representation. In H. Wellman (Ed.), Children’s Searching: The development of search skill and spatial representation (pp. 1-26). Hillsdale, NJ: Erlbaum. Pufall, P. B., & Shaw, R. E. (1973). Analysis of the development of children’s spatial reference systems. Cognitive Psychology, 5, 151-175.

Representation

39

Siegel. A. W., Herman, J. E, Allen, G. L., & Kirasic, K. C. (1979). The development of cognitive maps of large- and small-scale space. Child Development, 50, 582-585. Ungerer, J. A., Zelazo, P. R., Kearsley, R. B., & O’Leary, K. (1981). Developmental changes in the representation of objects in symbolic play from 18 to 34 months of age. Child D,?velopmenf, 52, 186-195. Vygotsky, L. S. (1978). Mind in sociery. Cambridge, MA: Harvard University Press. Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and constraining function of wrong beliefs in young children’s understanding of deception. Cognition, 13, 103-128. Winner, E. (1979). New names for old things: The emergence of metaphorical language. Journal of Child Language, 6 , 469-491.

This Page Intentionally Left Blank

CHILDREN’S UNDERSTANDING OF MENTAL PHENOMENA

David Estes DEPARTMENT OF H U M A N DEVELOPMENT A N D FAMILY STUDIES T H E PENNSY LVANlA STATE UNIVERSITY UNIVERSITY PARK, PENNSYLVANIA 16802

Henry M. Wellman and Jacqueline D. Woolley DEPARTMENT OF PSYCHOLOGY UNIVERSITY OF MICHIGAN A N N ARBOR, M I C H I G A N 48109

1. INTRODUCTION 11. CHILDHOOD REALISM 111. MENTAL ENTITIES

1V. PRlOR STUDIES V. CURRENT STUDIES A. EXPERIMENT I: CLQSE IMPOSTORS B. EXPERIMENT 2: CHILDREN’S UNDERSTANDING OF MENTAL IMAGES C. EXPERIMENT 3: MENTAL IMAGES VERSUS PHOMGRAPHS D. COMPARISON WITH LAURENDEAU AND PINARD

V1. GENERAL DISCUSSION VII. CONCLUSION REFERENCES

I.

Introduction

Adults possess an intricate set of conceptions about mental phenomena. They distinguish between thoughts and physical objects, between imagining something and doing it. They conceive of various mental processes, such as 41 ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR. VOL. 22

Copyright

G 1989 hy Academic Press, Inc.

All riglilr of reprodllctioii

in

any form reserved.

42

David Estes et al.

remembering, fantasizing, and inferring, and they conceive of various mental entities, such as ideas, dreams, and images. This set of conceptions forms a commonsense theory of mind that encompasses everyday knowledge about mental phenomena. Cognitive science goes beyond such everyday concepts to more rigorous and technical analyses, but all normal adults attain some naive understanding of the mind and employ a variety of mental concepts to understand their own and others’ behavior. That we do so is clear when we use such terms as remembec think, believe,guess, idea, dream, and imagine. What about young children? The traditional proposal (cf. Piaget, 1929) has been that before 6 or 7 years of age, children are incapable of distinguishing between mental and physical phenomena. Young children instead are said (1) to identify mental activity, such as thinking, with external behavior, such as talking or doing; (2) to believe that dreams are external, objective events; and (3) to believe that the mind refers merely to the external head (Broughton, 1978; Keil, 1979, p. 128; Misciones, Marvin, O’Brien, 8z Greenburg, 1978; Piaget, 1929, p. 43). The label childhood realism (Piaget, 1929) has come to refer to this general view. In contrast, we propose that even very young children, from around 2$ years on, make a fundamentally correct distinction between mental and physical phenomena. This early understanding provides a strong foundation for building further appropriate conceptions of the mental world. The research we present here provides evidence for deciding between these two proposals regarding young children’s understanding of mental phenomena. Whether children are ignorant or knowledgeable about the mind is a n important issue because an understanding of the mind plays a central role in several other aspects of development. An understanding of mental phenomena is fundamental to understanding the physical world because immaterial abstract things such as dreams contrast with and help delineate material objects. Understanding the mind is thus part of children’s developing ontological knowledge-their understanding of different types of “things” (Keil, 1979). An understanding of the mind is also fundamental to an understanding of the social world of self and others (Shantz, 1983). Commonsense adult understanding of human action is largely mentalistic; overt acts are seen as the products of such internal states as beliefs, desires, hopes, wishes, and misconceptions. Thus, children must come to distinguish between accidental and intended behavior, between wishes and reality, between plans and outcomes, between truth and deception. Finally, understanding the mind is necessary in order to know how to use it. That is, some degree of knowledge of the mental world seems to be a prerequisite for metacognitive development, which is arguably instrumental to developments in memory, learning, and reasoning (Brown, Bransford, Ferrara, & Campione, 1983; Flavell, 1979; Flavell & Wellman, 1977; Wellman, 1985). Some recognition of the basic distinction between internal mental phenomena and external physical and behavioral phenomena seems pre-

Understanding of Mental Phenomena

43

requisite to acquiring a commonsense theory of mind. Identifying when children recognize this distinction has therefore been our initial focus. Specifically, in the work presented here, we have examined children’s understanding of the subjective, nonphysical status of mental entities, and hence whether or in what way children might be realists. We therefore begin by clarifying these notions: childhood realism and mental entities. Then we present three studies of young children’s understanding of mental entities.

11. Childhood Realism The term childhood realism was coined by Piaget. It has been used widely as a description of young children’s thinking, meaning roughly an inability to distinguish mental from real, physical phenomena. Piaget’s discussion of childhood realism is difficult and complex; the term has been appropriated and expanded by others; and childhood realism is importantly different from realism as commonly discussed in philosophy. In philosophical discourse, one sense of realism refers to the view that the everyday objects of our perception (e.g., rocks, other people) exist independently of our thinking of them, occupying portions of an external space encompassing them and us. This sort of realism thus affirms the commonsense notion of a world of real objects presenting themselves for our perception. Realism in this sense can be contrasted with idealism, which encompasses a variety of views holding that we have no true knowledge of the existence of such a real world (because all we have access to are our own subjective perceptions), or that indeed such an objective world does not exist, and “external objects” are nothing but our own ideas. Realism in this sense appeals to a fundamental ontological distinction between ideas versus independently existing physical objects. In this distinction, physical objects are clearly not ideas. Physical objects are material, external, public, and objective, whereas ideas are mental, internal, private, and subjective. Talking about ideas as physical objects is simply anomalous or metaphorical (Keil, 1979). Even idealist philosophers acknowledge this everyday distinction as a starting point for their arguments; their claim, however, is that we are mistaken in our commonsense belief that physical objects have an objective existence. Realism and idealism thus both recognize a potent naive dualism between thought and objects. Childhood realism, in contrast, ignores such a distinction, or better put, it precedes this dualism. Childhood realists, so the claim goes, construe ideas as physical objects and have no comprehension of the basic duality between thoughts and things. The construct of childhood realism as used by Piaget and subsequently by others confounds two distinguishable possibilities. We term these possibilities ontological (childhood) realism and epistemological (childhood)

44

David Estes et al.

realism. Piaget failed to distinguish clearly between these two separable misconceptions because he believed young children to be afflicted with both. By ontological realism we mean a belief that mental phenomena are real physical phenomena, for example, that dreams are external pictures observable by others. As Keil(l979) expressed it, young children “apparently think that all things are types of physical objects” (p. 128). Piaget spoke directly about this sort of conceptual adualism and attributed a strong form of this confusion to young children: “The child cannot distinguish a real house, for example, from the concept or mental image or name of the house” (1929, p. 55). Epistemological realism concerns a different aspect of our mental life. The focus here is children’s conception of the acquisition of knowledge. For example, if ideas are literally physical things (the ontological issue) then we could acquire them (the epistemological issue) in the same way that we acquire physical things. That is, we might collect them from the available objects we find in the world, or they might seek us out and attach themselves to us like a stray dog, or we might even build them up out of other available things and materials. Although this example shows that ontological and epistemological realism can be intimately connected, they are distinct notions. One refers to the existential status of ideas; the other refers to the origins of ideas. The distinction between these two sorts of realism can be made clear by envisioning a person who is an ontological dualist-that is, who conceives of ideas and physical things as fundamentally different-but who is at the same time an epistemological realist. Such a person might firmly distinguish between thoughts and objects as types of things, but nonetheless believe that ideas are directly, physically caused by objects. For example, such persons might “proceed as though they believe objects to transmit, in a direct-line-of-sight fashion, faint copies of themselves, upon anyone who happens in the path of such objective knowledge. Within such a view, projectile firings from things themselves bombard and actively victimize individuals who function as passive recorders and simply bear the scars of information which has been embossed upon them” (Chandler & Boyes, 1982, p. 391). Epistemological realism of this sort is to be contrasted with epistemological constructivism. From a constructivist perspective, ideas and even perceptions d o not directly impress themselves on us. As a result, different persons can have very different subjective impressions, thoughts, and perceptions of the same situation. Piaget did not distinguish between the ontological and epistemological aspects of childhood realism and thus did not consider the possibility that children might be afflicted with one form of realism but not the other. We have emphasized this distinction because we think it unlikely that children are ontological realists, but they may still be epistemological realists. Given

Understanding of Mental Phenomena

45

such a possibility, an obvious first question to address is whether young children are ever ontological realists. We have focused on this question in our research.

111. Mental Entities In order to assess children’s understanding of the ontological distinction between ideas and things, we have questioned them about mental entities, We therefore need to define mental entities and specify their place among other mental and physical phenomena. Our aim is not to resolve deep and persistent philosophical controversies. We simply intend to clarify our terminology. An important preliminary point is that in everyday (and scientific) usage, we employ the language of the external physical world to talk about mental phenomena. Ideas are “rough” or “clear,” someone’s thinking is “sharp” or “fuzzy.” By analogy, just as the physical world includes physical entities (rocks and houses), physical events (rainstorms and recitals), and physical-chemical-biological activities or processes (respiration and combustion), so too the mental world is describable as encompassing mental entities (thoughts and beliefs), mental events (episodes of dreaming and instances of forgetting), and mental activities or processes (remembering, thinking, freeassociating). This imperfect analogy highlights some useful distinctions within the mental realm. Thus, a conversation about dreams might focus differentially on dreaming as a process (rapid eye movement sleep, etc.), on dreams as events (with a story line that unfolds in time), or on dream entities (the monster in a dream). Much talk about the mental world involves such mental verbs as remember; guess, believe, think, know, dream. Statements using such verbs typically identify one sort of mental activity out of many-for example, dreaming versus remembering-and identify some content. Thus, I dream about a dog I guess what you will say; I imagine a unicorn. We use the term mental entities to refer to the mental contents or products of such everyday mental activities and statements. Mental entities, therefore, include the dream-dog in the statement, “I had a dream about a large dog.” Mental entities (e.g., a dreamed dog) are distinctly different from corresponding real entities (a dog), and this difference is a central instance of the general difference between thought and object, mental and physical, mind and reality. As is clear in the preceding examples, mental entities are representational entities. They represent and refer to other things-their referents. A memory of a particular large black dog known in childhood is not the same as that dog, but it represents, or mentions, or refers to it. Thus, we need to say something about the representational relationship between thoughts and

46

David Estes et al.

things. Again, we do not aim to resolve thorny issues about reference. We wish only to distinguish three separable possibilities. First, the referent of a mental entity may really exist (or have once existed), as happens when we think about an actual person. Or, the referent may be hypothetical-perhaps existing and perhaps not-as, for example, when a bachelor thinks of his asyet-unmet future wife. Finally, the referent may be mythical-clearly nonexistent or fictional-as, for example, when we think about a unicorn. In all these cases, however, the mental entity is mental and not physical or real (in one sense of that word). It is important to note that mental entities are not uniquely characterized by being representational. Photographs and drawings, for example, are real, concrete, physical objects (not mental entities) but they are also representations. Finally, a distinction can be drawn between a particular mental entity and mental representations generically. This is the distinction bet ween a thoughtabout-a-dog and thoughts, between a particular idea and ideas in general. Thoughts, ideas, dreams, and so on, in unspecified general terms have been termed “abstract objects” (Keil, 1979). In our research, we have questioned children about specific mental entities, but we can use their responses across a variety of instances as evidence of their understanding of abstract objects more generally.

IV. Prior Studies Our studies of children’s understanding of mental entities began with an intuitive analysis of the criteria that adults invoke to distinguish between mental entities and real physical objects. Initially, we identified three such criteria (Wellman, 1985). First, real objects, as opposed to their corresponding mental entities, afford behavioral-sensory evidence. You can see and touch a chair but you cannot do these things to its mental representation, for example, a thought about a chair. Second, real physical objects have a public existence. Other people can see them and touch them. In contrast, a dreamedof entity might seem as if it can be seen and touched, but no one else experiences it similarly. Third, a real physical object has a characteristic temporally consistent existence. Mental entities do not have the same sort of consistency. Images can come and go simply by willing them. People can have different dreams each night, but the waking world is more consistent. We used these reality criteria to test children’s understanding of the distinction between physical objects and mental entities. In our first study (Wellman & Estes, 1986), we presented 3-, 4-, and 5-year-old children a number of contrasts between two characters. For example, one presentation was a description of a boy who had a dog and another boy who was thinking about a

Understanding of Mental Phenomena

41

dog. Other contrasts involved characters who were dreaming, remembering, or pretending. The child was asked to make judgments about each character’s “dog” on the basis of the three reality criteria, that is, to judge whether it could be seen, touched, and petted (behavioral-sensory evidence); whether someone else could see it (publicness); and whether the “dog” would be consistently available in the future (consistency). Even 3-year-olds were correct in ascribing behavioral-sensory, public, and consistent status only to real objects 72% of the time-greatly in excess of chance. When children did make errors they did not display any systematic tendency to interpret mental entities as objective and real; they were as likely to ascribe mental (not-real) status to real items as they were to ascribe real status to mental items. However, perhaps the children tended to misinterpret the presentations about mental entities to be specifying merely nonpossession. That is, they may have incorrectly interpreted the mental terms we used not as specifying a mental experience, but rather as specifying that the focal character simply did not have the object. “This boy is thinking about a cookie” thus may have merely meant to the child, “This boy hasn’t got a cookie.” If young children tended to interpret mental verbs in this fashion, then they would have tended to respond correctly, but for realistic reasons. Such a misinterpretation would constitute a subtle but definite form of realism-construing mental entities as real but absent physical objects. In a second experiment (Wellman & Estes, 1986), we again described mental and real entities and had children judge them on the basis of the three reality criteria. Four kinds of presentations were used: (1) a character having a mental experience, (2) a character having a mental experience, with an explicit mention of nonpossession of the real object, (3) a character possessing a real object, and (4) a character not possessing a real object. If mental stories tend to be misinterpreted as simple nonpossession stories, such misinterpretations might happen more frequently when explicit mention of nonpossession is made. In this experiment, we also asked children to provide explanations for their answers. Once again, correct responses were very frequent in all three age groupseven 3-year-olds were correct 87% of the time. Children performed identically on simple mental items and on mental explicit nonpossession items. Finding that explicit mention of nonpossession did not enhance children’s answers to mental items suggests that children correctly distinguished mental entities from nonpossessed ones. More definitive evidence of such correct distinctions was available in children’s explanations. When asked if a nonpossessed apple could be eaten, children said no, because “it’s not there” or “it’s gone away.” They thus gave location-possession explanations for nonpossessed items. When asked if a thought about an apple could be eaten children also said no, but they typically explained this answer by saying that the entity was

48

David Estes et al.

distinctly mental (e.g., “it’s just imagination”), or they contrasted its reality status with that of a physical object (e.g., “it’s not a real apple”).

V. Current Studies The foregoing results make clear that children are not ontological realists in the strong sense of equating mental entities with their corresponding physical referents. Thus, Piaget’s (1929) claim that “the child cannot distinguish a real house. . . from the concept or mental image or name of the house” (p. 55) seems an overstatement. The possibility remains, however, that children might hold some more subtle realistic misconception. Wellman and Estes (1986) provided evidence against one such misconception-construing of mental entities as absent or nonpossessed physical entities. Other sorts of realistic misconceptions are clearly possible, however. For example, ideas and dreams may indeed be conceived of as physical things, but as special, seemingly insubstantial types of physical things, perhaps equivalent to naive conceptions of such real substances as smoke or shadows or air. Piaget (1929) in fact, quoted older preschoolers as saying that thoughts are smoke, air, shadows, lights, or sounds. A number of less substantial physical entities of this sort exist. For adults, however, mental entities contrast with even these “close impostors,” as we call them. The following list provides some contrasting examples. 1. Real external objects: material, public, external things 2. Absent real objects: like real external objects, but outside of sensory range 3. Insubstantial objects (e.g., air, smoke): like real external objects, but seemingly intangible and/or invisible. 4. Physical representations (e.g., drawings, pictures): material, public objects but representing some other object with very different physical features. 5. Internal sensations (e.g., pain): private and internal but localizable and still physically real experiences. 6. Mental entities (e.g., ideas, dreams): immaterial, private, and symbolic cognitive representations. Given the subtle features of mental entities (Item 6), a young child might plausibly identify them with something like smoke, shadows, pains, or pictures (Items 3-5) even if he or she does not identify them with prototypic external objects (1) or with absent objects (2). The previously listed sorts of entities populate one region of a larger conceptual space. This space is an ontological one, encompassing different kinds of possible things. We do not believe (as claimed by Keil, 1979)that ontological

Understanding of Mental Phenomena

49

knowledge is strictly hierarchically organized. But this space can be partitioned reasonably, and different sorts of entities can be contrasted with one another. We are generally interested in children’s understanding of this larger conceptual space, but our specific interest is in whether children recognize one basic division within this space-that between mental (Item 6) and physical (Items 1-4) entities. A. EXPERIMENT 1: CLOSE IMPOSTORS

To see whether young children differentiate mental entities from close impostors, we posed children simple judgment questions like those used by Wellman and Estes (1986), but we were interested primarily in their explanations for their judgments. We especially wished to examine whether children distinguish in their explanations between mental and physical items designed to elicit identical judgments regarding various physical acts. One cannot grab the shadow of a tree, nor can one grab a dream tree, but adults explain these physical impossibilities quite differently.

I. Method a. Subjects. Fifty-nine preschool children were tested: 19 3-year-olds (mean age 3 years, 4 months; range 2:11 to 3:11), 20 4-year-olds (mean 45; range 4:O to 4:11), and 20 5-year-olds (mean 5:9; range 5:3 to 6:6); 32 were girls and 27 were boys. The children attended a racially mixed but largely white middle class preschool.

b. Procedures. Each child heard 18 brief stories of two to three sentences each and responded to three behavioral-sensory questions after each story. Each story was accompanied by a simple black and white line drawing of a child character who was a boy in nine of the stories and a girl in nine. The objects in the stories were not depicted in the drawings. Three general types of stories were used, involving mental entities, solid physical objects, and close impostor entities. All the entities are shown in Table I. The alignment of the rows in this table shows our attempt to include comparable contents across the different item types. Thus, every mental x can be contrasted with an identical solid physical x , and many of the solid physical items can be contrasted with close impostor items. An example of the stories is, “See this boy. He always wanted a bicycle. Sometimes when he is asleep, he dreams about a bicycle.’’ Following each story, the child was asked whether the story character (1) could see the item referred to in the story, (2) could touch the item, and (3) could hide the item under his or her bed. Each question was prefaced by a simple clause that

David Estes et al,

50

TABLE I Items Used in Experiment lQ Mental entities I. Thought: Ball _--__-_2. Dream: Bike 3. Memory: Baby food - - - -

Solid physical objects 4. 5. 6. 7. 8. 9. 10. 11.

Real ball Real bike Real baby food Real tree ----- Real toothpaste - - Real lion -----Real leaves - - - - Real sand - - - - - -

Close impostors

12. Shadow of a tree 13. Used-up toothpaste 14. Photograph of a lion 15. Burned-up leaves 16. Smoke 17. Pain: Tummy-ache 18. Sound: Beep

“Dashes indicate comparable content across item types.

reiterated the gist of the story (e.g., ”Now that this boy is dreaming about a bicycle can he touch that bicycle with his hands?”). Explanations for all answers were solicited by asking, “Why cadcan’t the boy/girl see it with hidher eyes/touch it with hidher hand/hide it under hidher bed?” Questions were always asked in this order. The 18 items were presented in two randomly determined orders.

2. Results Children’s explanations were our primary interest, because yes/no answers to the questions asked often do not discriminate close impostors from mental items. Nevertheless, certain yes/no judgments do differentiate between mental and solid physical entities. These judgments provide validation for our procedures and preliminary evidence regarding children’s understanding of this distinction.

a. Judgments. We compared children’s judgments on the mental versus solid physical items for the see and touch questions only. Comparisons of these questions for these items replicate comparisons reported by Wellman and Estes (1986). Even the youngest children therefore should judge that mental items fail such behavioral-sensory tests and solid physical objects pass them. This expectation was confirmed-children at all three ages judged mental entities differently from solid physical objects. When the children were asked if someone could see or touch a physical object, 90% of their responses were yes, as appropriate. When asked if mental entities could be seen or

Understanding of Menial Phenomena

51

touched, 75% of the responses were appropriately no. The relevant percentages are presented in Table I1 for each age group. Four 3-year-olds were not included in these data because they indiscriminately answered yes to every question (even to such questions as can the boy hide a real, large tree under his bed). The mean age of the remaining 3-year-olds was 3:6. A 3 (Age) x 2 (Item type: mental versus physical) x 2 (Question type: see versus touch) analysis of variance revealed the expected main effect of item type, F(1, 52) = 391.19, p < .0001. Several interactions involving age were also significant, but all were subsumed under a significant three-way interaction, F(2, 52) = 8.18, p < .001. Separate Item x Question analyses of variance at the three ages clarified this interaction. For 3-year-olds and 4-year-olds there was only an effect of Item: F(1, 14) = 30.35, p < .001, and F(1, 19) = 530.40, p < .001, respectively. For 5-year-olds, however, both Item and Question effects were significant and subsumed under a significant Item x Question interaction: F(1, 19) = 19.32, p < .001. In short, all children appropriately judged mental items differently from physical ones. In addition, 3- and 4-year-olds responded similarly to the see and touch questions; 5-year-olds responded differentially to these questions, frequently claiming that mental entities can be seen.

b. Explanations. Children’s judgments demonstrated their basic understanding of the behavioral-sensory differences between mental entities and solid physical ones. We turn now to their explanations. Of special importance are children’s explanations for mental entities versus close impostors. Although these two types of entities often elicit similar yes/no judgments, most adults would support such judgments with very different explanations. Across children (n = 5 9 , items (18), and questions (3) we collected 2970 explanations. Children’s explanations typically fell into one of four superordinate substantive categories, or into a residual category including “I don’t know,” incomprehensible replies, and refusals to answer. The four categories were (1) mental descriptions (e.g., “He’s just imagining it”), (2) reality status considerations (e.g., “It’s not real”), (3) location-possession explanations (e.g., TABLE I 1 Proportion of Positive Responses to Behavioral-Sensory Questions in Experiment I 3-Year-olds Question See? Touch?

4-Year-olds

5-Year-olds

Mental

Physical

Mental

Physical

Mental

Physical

.41

.86 .71

.I5 .I1

.93 .93

.42

.36

.05

.98 .96

52

David Estes et al.

“It’s not there”), and (4)physical ability-moral constraint considerations (e.g., “It’s too big,” or “Her mom doesn’t let her”). This categorization scheme is presented in detail in Table 111. These four categories captured 82% of the responses. Ten percent of all explanations were coded by two independent coders. Intercoder reliability (calculated as agreements divided by agreements plus disagreements) was .90. Our central question concerned children’s understanding that, for example, although neither a dream nor smoke can be touched, dreams and smoke still differ in important respects. We wish to show that children recognize mental entities to be different from other kinds of entities which also fail simple behavioral-sensory tests. We therefore focused our analyses on children’s explanations of their negative judgments (that something can not be seen, touched, or hidden). Negative judgments overall made up 62% of the total judgments, and 77% of all judgments of mental entities and close impostors. Star graphs (Anderson, 1960) provide an informative means of representing children’s explanations. Nine different star graphs are presented in Fig. 1. Each of the four arms of each star corresponds to one category of explanation. Plotted along each arm is the percentage of total explanations (from the four main categories) of that one type. The plots along each arm are then connected to form a graphic figure.

TABLE 111 Categories of Explanations in Collapsed Coding Scheme 1. Mental descriptions: Child uses a mental term to explain his or her judgment. Mental terms

were noted to indicate if they were new (i.e., not in the presentation), and if they were elaborated or qualified in some way-for example, “He’s thinking about it in his head” (elaboration) or “He’s just dreaming” (qualification). 2. Reality sfafusconsideralions: Child appeals to behavioral-sensory reality criteria (e.g., “She can see it”; “It’s invisible”) or makes an explicit statement about the reality status of the item (e.g., “It’s real”; “It’s not true”). 3 . Location-possession: Child refers to the location of the itern or its possession by the character (e.g., “It’s there”; “She has one”). 4. Physical ability-moral constraint: Child refers to some physical property of the item (e.g., “It’s too big”), a bodily consequence (e.g., “It would burn her”), a property consequence

(e.g., “It would mess up the rug”), organ capacities (e.g., “She has eyes”), or moral proscriptions (e.g., “Her dad will be mad”). 5. Incomprehensible-don’t know: Uncodable responses were those that did not fit into any of

the preceding codes, and usually appeared to be a result of the child’s being sidetracked. This category, in addition to explicit “I don’t know” responses, includes instances in which no response at all was given.

53

Understanding of Mental Phenomena

Solid

Phvsical I t e m s

Mental

Close IrnDostors

Items

-

-

-,3-Year-Olds 100%

100%

100%

I

I

I

Considerations Reality 0

0

’

possession Location-

-

!\

4-Year-Olds 100% \

I

7/

- 100%

A

100%

100%

-.

1

Reality 0 Considerations

1

0

-

1

-

-

100%

100%

I/{ Locationpossession &

-

5-Year-OldS 100%

100%

100%

I

I

I

Reality 0 Considerations

0”

100%

possession

Physical abilityMoral constraint

Physical abilityMoral constraint

Physical abilityMoral constraint

Fig. 1. Star graphs for childrenk explanations in Experiment 1. Each star represents the percentages of the four d g e e n t exphnation types given for a single item type by one ofthe three age groups.

Figure 1 shows children’s explanations for their negative judgments on the three item types. Examination of these graphs reveals first and foremost that children at all ages used distinctively different patterns of explanation for the different item types. When children said that solid physical items cannot be seen, touched, or hidden, it was for physical or moral reasons. In most cases, the negative judgments for solid physical items occurred for the “hide under the bed” question. For example, It can’t be hidden under the bed, “because it’s messy” (babyfood), “because it’s too big” (bike), or “because

54

David Estes et al.

her mom will get mad” (toothpaste). Comparing children’s explanations for these physical objects and for mental entities shows that young children have different conceptions of mental and physical entities. In contrast to the physical-moral explanations given for solid physical items, mental items elicited predominantly mental explanations. Even for 3-year-olds, mental items elicited more mental explanations and fewer physical explanations than did physical items. Moreover, explanations for mental entities were very different from explanations for close impostor items. Four Age x Item-type analyses of variance were conducted on these data. Each analysis focused on one of the four explanation types: mental, physicalmoral, reality-status, and location-possession. Four separate analyses were conducted because the explanation types are partly dependent (i.e., total explanations across these four categories sum to 100%). These analyses confirmed the graphic depiction of the results in Fig. 1. At all ages mental explanations were most often provided for mental entities-evident in a main effect of item type in the analysis of mental explanations, F(2, 104) = 75.41, p < 001. In contrast, physical-moral explanations predominated for solid physical entities, F(2, 104) = 72.00, p c 0001. And location-possession explanations were most frequent for close impostor entities, F(2, 104) = 65.03, p < .0001. For mental explanations and for physical-moral explanations, the main effects of item type were subsumed under Age x Item-type interactions: < .001,and F(4, 104) = 2.80,~< .05, respectively. The F(4, 104) = 7.97,~ interaction for mental explanations indicates that mental explanations increased with age, but, appropriately, this increase was confined to explanations for mental items. Thus, simple effects showed nonsignificant age differences in mental explanations for physical items and for close impostor items, but significant increases with age in mental explanations for mental items: F(2, 52) = 6.54, p < .01. The interaction for physical-moral explanations indicates that mental entities rarely received physical-moral explanations from 4- and 5-year-olds but did receive such explanations from 3-year-olds. Thus, the simple effect of Age for physical-moral explanations on mental entity questions was significant: F(2, 52) = 9.04,p < .001.Nonetheless, 3-yearolds never gave as many physical-moral explanations for mental entities as they did for solid-physical entities; thus the simple effect of Item type was significant even for 3-year-olds: F(1, 14) = 7.01, p < .02. The quantitative data do not provide a good sense of the appropriateness of these young children’s explanations. Because we are particularly interested in children’s mental explanations for mental items, Table IV presents some examples of such explanations. Forty-seven percent of the 3-year-olds, 80% of the 4-year-olds, and 100% of the 5-year-olds gave mental explanations. Of the mental explanations given, 20% simply but appropriately repeated the mental term used in the story presentation (e.g., in answer to why a boy

Understanding of Mental Phenomena

55

thinking of a ball could not touch it, “He’s just thinking of it”). Table IV presents five mental explanations at each age, randomly sampled from those in which the child did not just reiterate the presented mental term. This random sample gives a flavor of children’s typical mental explanations. In addition, Table IV presents three additional explanations at each age, chosen to provide an indication of the sophisticated responses that many children occasionally achieved.

3. Discussion Two aspects of the results from Experiment 1 deserve discussion: (1) the evidence that children at all ages appropriately distinguish mental entities from even subtle close impostors such as smoke, shadows, pictures, and sounds,

TABLE I V Mental Explanations from Experiment 1 Randomly sampled mental explanations’

Other informative explanations*

~

3-Year-olds ’Cause it’s her imagination. Because he’s just remembering it in his brain. ’Cause it’s in his mind. Because he’s thinking it and he can’t see it. Because it’s in his imagination.

3-Year-olds It’s just in his head: he’s imagining a bicycle; a dream of a bicycle. Because he is only imagining it. Because he’s just dreaming about it.

4-Year-olds Because it’s in his head. ’Cause it’s just baby food and he’s not eating it, he’s thinking about it. ’Cause it’s in his mind. ’Cause he’s just thinking about it. ’Cause her hand can’t fit in her head.

4-Year-olds No, because it’s only a dream; but he could touch it with his mind hands. ’Cause it’s invisible; ’cause you just imagine it. ’Cause it’s inside his head, he’s just thinking about it.

5-Year-olds Because it’s imaginary. Because it’s not there; it’s in his head. Not real; it’s in his mind. Because it’s not real; he just sees it in his imagination.

5-Year-olds Because if you’re blind or not you can still see things in your imagination. Because it’s in his head, or in his eyes, or wherever it is. Because he’s just looking at it in his imagination.

’Five randomly sampled mental explanations that do not simply repeat the mental verb used in the story. *Three mental explanations, from three different children, that seem especially clear or informative.

56

David Estes et al.

and (2) the findings indicating some changes in age in the nature of children’s responses. a. Against Childhood Realism. At all ages, children appropriately distinguished between mental and physical entities-even invisible and intangible physical entities. These results indicate that children as young as 3 are aware of the mental realm, and are aware of some of the basic differences that distinguish it from the external physical world. Clearly, our data agree with those of Wellman and Estes (1986) in showing that young children are not realists in the sense of equating mental entities with the physical objects to which they refer. More important, our data argue that young children also are not realists in the more subtle sense of confusing mental entities with physical but intangible entities such as smoke, shadows, or sounds; nor d o they confuse mental entities with physical representational entities such as photographs. Our hypothesis was that if children did pass through a stage of subtle realism in which mental entities were confused with things like shadows or air, then during this stage they should say that mental entities have the same properties as these close impostors. No evidence from our research supports this view. In this respect, even 3-year-olds demonstrated some knowledge of the categorically mental quality of mental entities. An important methodological issue concerns the differences between the methods we have used and those used by prior researchers such as Piaget (1929) and Laurendeau and Pinard (1962). According to Laurendeau and Pinard, the methods used to investigate the sort of phenomena under study here can be divided into two types, which accordingly lead to divergent conclusions. Piaget’s method of analysis involved a global evaluation of all the child’s answers to a group of questions-a subject-focused holistic approach. The other type, represented for example by the Wellman and Estes (1986) studies, is more analytic. It involves a n item-centered analysis, determining for each question the frequency of certain types of responses. Although both methods have advantages, Laurendeau and Pinard strongly favored the former. They found two flaws in the latter method. First, in the item-centered method, the reasons given by the child in justification of his or her answers are typically neglected. Second, this method is focused on discrete items independently from one another, thus neglecting the child’s attitude toward the questions and the topic as a whole. The methods employed in the current study circumvent these objections. We made children’s explanations the focus of our analysis, yet the content of children’s explanations still shows that they appropriately distinguish between mental and physical entities. Regarding the second criticism, we too were interested in capturing an overall profile of children’s conceptions, integrated across various individual items. Star graphs were used for this

Understanding of Mental Phenomena

51

purpose. These figures graphically capture a global characterization of how children of a given age talk about a particular category of items, and they do so in a way that portrays the relationships among different types of items.

b. Developmental Differences. Our data also show that 3-year-olds, while significantly distinguishing mental entities from physical ones, do not do so as consistently as 4- or 5-year-olds. In particular, 3-year-olds often used physical-moral explanations (“You can’t touch it because you’ll get too dirty”) to explain their judgments of mental entities. One possible explanation of these findings is that such young children are displaying remnants of a realistic conception of mental entities. If such a description characterizes 3-year-olds’ then perhaps children younger than 3 are firm realists and the basic developmental sequence from ontological realism to subjectivism, as described by Piaget, is correct but occurs earlier in childhood than he believed. We favor an alternate interpretation of 3-year-olds’ errors, however. When this kind of error occurred, children’s comments often suggested that they were talking about the object referred to (a thought about a ball), not the mental entity itself (a thought about a ball). Mental entities are representational entities and thus involve both a representation (a thought) and a referent (about a ball). Young children at times may simply confuse representation and referent, replying to questions about the representation with answers about the referent. Often questions about a person’s thoughts (“Tell me about your dream”) are requests to talk about referents (“There was a big white dog coming to get me”). The developmental trend may indicate that 3-year-olds, while largely correct, are more likely than older children to answer our questions by talking about the referent rather than the mental entity. We have tried to address this issue and to provide a richer, more comprehensive view of children’s understanding of the mental realm by conducting two studies of children’s conception of another mental phenomenon, mental imagery. B. EXPERIMENT 2: CHILDREN’S UNDERSTANDING OF MENTAL IMAGES

Studying children’s understanding of mental images allows us to address several important issues. 1. In our prior studies, we have had children make judgments about the mental entities of another person. Examining their conceptions of their own mental entities, such as their own mental images, would extend our findings from others to self. 2. This extension from others to self is related to another important extension. Mental entities have both a conceptual and an experiential aspect.

58

David Esres et al.

Conceptually, one acquires knowledge of mental entities in general and forges an understanding of how they differ from physical entities. Experientially, one experiences specific mental phenomena. We apprehend mental entities as first-hand, subjective experiences. Indeed, the conscious “qualia” of our mental states are sometimes claimed to be their primary distinctive feature, that aspect of mind that sets human thought apart from computer “thought,” for example (Dennett, 1979, Chapter 9). Not surprisingly, then, Piaget’s description of young children’s deficient understanding of mind also had both a conceptual and an experiential aspect. Ontological realism, as described earlier, concerns children’s putative conceptual misunderstanding. Experientially, young children were also supposed to be incapable of introspection. According to Piaget’s account, young children are unable to reflect on their own mental processes and have no awareness of the boundary between the internal mental world and external reality. The ability to introspect in this sense (i.e., the simple capacity to be aware of having thoughts, which all normal adults possess) can be distinguished from conscious access to functionally significant cognitive processes, an issue of considerable debate in cognitive science (Ericsson & Simon, 1980; Nisbett & Wilson, 1977). Our concern, and Piaget’s, is simply whether young children have the capacity to be aware of their own thoughts. Piaget (1928) first claimed that young children are incapable of introspection while investigating the development of mathematical ability. He found that children could not reconstruct the process they used to solve simple arithmetic problems. When asked to do so, they related a series of more or less arbitrary steps that had little or nothing to d o with the problem in question. Here is an example involving a 7-year-old: Adult: This table is 4 meters long. This one is three times as long. How many meters long is it? Child: 12 meters. Adult: How did you do that? Child: I added 2 and 2 and 2 and 2 and 2 always 2. Adult: Why did you take 2? Child: So as not to take another number. (Piaget, 1928, p. 139)

Piaget inferred from such responses that introspection is completely absent until at least 7 or 8 years of age, concluding that young children d o not have the ability to step back and take their own mental states and processes as objects of reflection and discussion. Piaget claimed that children only slowly gain introspective access to their own thoughts: the process of “thought becoming aware of itself’ is a gradual one that is not completed (i.e., does not reach the normal adult level) until the very end of childhood. If children did lack reflective awareness of their own thoughts, then this limitation would

Understanding of Mental Phenomena

59

severely interfere with their understanding of mental phenomena. Their understanding of mental phenomena would surely be much different and more impoverished than ours, enriched as it is by introspective access to our own mental states. Examining children’s understanding of their mental images is a useful way to address this issue. 3. Given our prior findings, which are inconsistent with ontological realism, we wanted to devise a strong, perhaps decisive test of that position. Mental images are a useful vehicle for such a test. Mental images share several properties with concrete pictures or photographs. Thus, if children were going to confuse some mental entity with a physical object, a confusion between mental images and photographs is a very plausible one. Further, in our mental imagery studies, we termed mental images “pictures in your head” when talking about them to children, This terminology was intended to be an open invitation for children to indulge in mistaken realistic conceptions. Such phrasing encourages the child to conceive of images literally as real physical pictures. We hoped to show that children’s understanding of the crucial distinction between physical objects and mental entities is clear and compelling even in a case that seems ripe for realistic confusions. 4. Finally, we wanted to probe further into children’s understanding of the distinctive features and properties of mental phenomena. In this regard, the judgments made by children of mental entities so far are consistently negative. Mental entities are not real, can not be touched, can not be seen, etc. But mental phenomena have important positive features as well. For example, one can imagine a mental light bulb turning on simply by thinking of it doing so. Simply thinking will not turn on a real light bulb. Mental entities, unlike physical entities, can be manipulated and transformed by mental effort and attention alone. Are children aware of this property? Focusing on the transformational properties of mental images has methodological advantage as well. In querying children about their ability to transform mental images, we devised a question about mental entities that required a positive response. This procedure allowed us to rule out a simple negative response bias as an explanation for some of our prior results. The upper left portion of Fig. 2 shows how we incorporated all these aims into the design of our first study on images. The basic design is depicted by showing an idealized pattern of correct responses to questions about three contrasting sorts of entities: (1) real objects, such as a deflated balloon; (2) real hidden objects, which are currently invisible and inaccessible, such as a deflated balloon hidden in a box, and (3) mental images of real objects, such as a mental image of a deflated balloon. For the mental image questions, we first showed the children the balloon, asked them to close their eyes and make a “picture of it in your head,” and then questioned them about

;I-/

David Estes et al.

60

3-YEAR-OLDS

(Yes)’!:

25

\Object \

......*\Hiddenobject

No

See? Touch? Other Tmnsform? See?

4-Y EAR-OLDS

---

(No)

.....4

Ima

...........-..-......................

Hidden Object See?

Touch? Other See 7

Tronsform?

5-YEAR-OLDS

------

..................................... See?

Touch? Other Transform? See?

............

.......

”

See? Touch 7 Other See?

Transform?

Fig. 2. Responses to the four questions for each type of entity in Experiment2. The top left graph shows an idealizedpattern of correct responses. The othergraphsshow actual performance by each age group.

“that balloon in your head.” The questions we asked are shown along the horizontal axis in Fig. 2. They were “That balloon in your head, can you see it with your eyes?” and “Can you touch it with your hands?” (behavioralsensory questions). “Can I see it with my eyes?” (a publicness question). “Just by thinking about it, can you make it stretch out long and skinny?” (a transformation question). This design permits evaluation of several alternative hypotheses, each of which would result in patterns of response different from the ideal one. If children mistakenly think that images are real external objects like pictures, for example, they should answer about images just as they do about visible objects. More plausibly, if children think that images are real but inaccessible objects-literally pictures inside the head-then they should answer about images as they d o about hidden objects. Finally, if children simply have a general negative response bias with respect to questions about mental entities, they would respond negatively to all image questions.

Understanding of Mental Phenomena

61

The correct response pattern is labeled an idealized one, and it is simplisticallyidealized in one respect. We would not necessarily expect children (or adults, for that matter) to deny consistently that they see their images, because “seeing” an image seems an acceptable if imprecise way of talking. Note, however, that we phrased the see question as, “Can you see (the item) with your eyes?” Given that phrasing, we did expect children to affirm less often that they see their images with their eyes than that they see real objects with their eyes.

1. Method a. Subjects. A total of 72 preschoolers participated: 24 3-year-olds (mean 3:7; range 3:l to 4:O); 24 4-year-olds (mean 47; range 4:l to 49); and 24 5-yearolds (mean 5 5 ; range 5 0 to 5:lO); 33 were girls. The children attended the same preschool described in Experiment 1. b. Design. As is clear in the top left portion of Fig. 2, the transformation question is crucial to the logic of this design. That mental images but not physical objects can be transformed mentally creates the desired case in which a positive response is correct for a mental entity and a negative response is correct for a physical object. Because of its centrality to the logic of the design, we had to have a sensitive measure of children’s knowledge of this property of mental images. One problem was the difficulty in being certain that the question about transformational imagery was unambiguous for preschool children. Because transforming a mental image is an activity or process, however, we were able to ask not only whether it can be done, but also that it actually be attempted. Thus, if children at first claimed to be unable to transform a mental image, they were asked to try to do so and were queried as to their success or failure. Of course, probing children in this fashion might simply make them change their original response (without trying to or succeeding in transforming the mental image), so an appropriate control was necessary. The same probe was therefore used after some responses to transformation questions on hidden object items, allowing assessment of whether the probe itself induced children to change their initial answers.

c. Procedure. We began with a warm-up task to familiarize children with the terminology used to talk about mental images. Children were shown a drawing of a crescent with upwardly oriented ends and were told, “Close your eyes and think about this happy smile. Try to make a picture of that happy smile in your head.” After claiming they had done so, they opened their eyes and watched as the picture was rotated 180 degrees. The experimenter said, “See how I can make it turn into a sad frown?” Then they were told, “Think real hard and try to turn that happy smile upside down in your head until

62

David Estes et al.

it’s a sad frown.” Six children (four 3-year-olds, one 4-year-old, and one 5-yearold) claimed to be unable to form a mental image in response to these instructions and were therefore excluded from further testing. For the primary task, three objects were used-a cup, a pair of scissors, and an uninflated balloon. Each child was thus questioned about nine entities-a mental image of each of the three objects and the objects themselves when visible and when hidden. These entities were all referred to with parallel terminology as “the (object) in your head,” “the (object) on the table,” and “the (object) under the box,” respectively. On mental image presentations, after the object was shown to the child and he or she had been asked to form a mental image of it, it was placed out of sight in a box on the floor. The four questions in Fig. 2 were asked about all nine entities. In its general form, the transformation question was, “just by thinking real hard, without moving your hands, can you -?” The transformations were, for the cup, “make it turn upside down,” for the scissors, “make them open and close,” and for the balloon, “make it long and skinny.” When children denied being able to transform a mental image, the probe procedure was used: The child was asked to try to form the mental image (“make a picture of the cup in your head”) and then to try to make the transformation. The child was then asked whether he or she had made the transformation. This probe procedure was also used after at least one hidden object transformation question per child. The questions for each item were asked in one of two counterbalanced orders. The entities were presented in one of two blocked orders, with the block of mental image items presented at either the beginning or end of the session. Within the block of object items in each of these two orders, half the children were initially questioned about each object when it was visible and then were questioned about it after they had seen it placed under the box; the other half were initially questioned about each object after they had seen it placed under the box and then were questioned about it after the box was removed. Question order and item order were counterbalanced across children. Explanations were requested (with “Why?” or “Why not?” as appropriate) for all responses on two of the image items and on two of the hidden object items for the 4-and 5-year-olds and on one item of both kinds for the 3-year-olds.

2. Results a. Judgments. Preliminary analyses revealed no effects of item order, question order, gender, or object (balloon, cup, or scissors), and no interaction among these variables. Figure 2 indicates the correspondence of each age group’s performance to the ideal pattern. The obvious primary conclusion from these data is that although overall correspondence to the ideal

Understanding of Mentd Phenomena

63

pattern improved somewhat with age, even the 3-year-olds’ answers corresponded, in essence, to the ideal pattern. These young children judged mental images differently from both visible and hidden objects. Most important, all age groups exhibited the appropriate cross-over from negative to positive responses on the image transformation question and from positive to negative responses for the visible object transformation question. As expected, the largest deviation from the ideal.pattern was on the see question for mental images. Finally, as in Experiment 1, in spite of generally good performance, the younger children did at times answer that they could touch their mental images and that they could be seen by someone else. Statistical analyses confirmed what is obvious in Fig. 2. A 3 (Age) x 3 (Entity type) x 4 (Question) analysis of variance on correct responses revealed main effects of age and entity type as well as several interactions, all subsumed under an Age x Entity x Question interaction: F(12, 414) = 2.72, p < .01. This three-way interaction captures the apparent deviations from the ideal pattern as depicted in Fig. 2. Specifically, the interaction is largely explained (1) by the performance of the 5-year-olds on the visible and hidden object transformation questions, which was better than that of the 3-yearolds (ts(46) > 2.76, ps < .01) and the 4-year-olds (ts(46) = 2.5, ps < .05), and (2) by the 4-year-olds’ performance on the question about seeing the image, which was considerably more deviant than performance on any other question by any age group (ts(23) > 1.58, p c .05). Given these deviations from the ideal pattern, it is important to emphasize the consistent conformance by all age groups to that pattern: Except for the performance of the 4-year-olds on the question about seeing the image, mean correct performance by all age groups on all questions was better than a chance level of 50%, (ts(23) > 3.14, p < .005). These analyses and the data in Fig. 2 are based on children’s final response to the transformation questions when the probe was used. Figure 3 compares initial and final answers to the transformation question (before and after the probe) for both mental images and hidden objects. Consider the mental image items first. If initial responses alone are considered, only the 5-year-olds consistently say that they can transform mental images (mean = 82%, t(23) = 5.63, p < .Ol); 3- and 4-year-olds are responding at chance (mean = 51% and 53%, respectively, t s c 1). However, when they were asked to try to transform a mental image (after initially claiming to be unable to do so), performance by each age group improved, and dramatically so for the 3- and 4-year-olds, ts(23) > 3.12, ps c .005. When final responses are taken into account, performance approached ceiling and no age differences were obtained (93V0, 93%, and 96% yes responses for the 3-, 4-, and 5-yearolds, respectively). The pattern of initial to final responses on the transformation question

David Estes et at

64

YES JUDGMENTS ON MENTAL TRANSFORMATION QUESTIONS 100 (YES) yo

-

75-

50-

4

53

7

25t “0)

0 ye

I I Initial Final MENTAL IMAGE

-3 4

=\ I

I

Initial Final HIDDEN OBJECT

Fig 3. Performance on the transformation questions in Experiment 2 both before (initial)and after flnal, the probe question.

for hidden objects is shown in the right half of Fig. 3. In this case, children rarely changed their answers, and initial and final responses did not differ for any age group. Thus, these children were not simply changing their answers in response to probed questioning by an adult.

6. Individual Analyses. Individual performance was consistent with the results of these group analyses. Each child was asked a total of 12 questions on the three mental image items. The probability of answering correctly by chance 9 or more of 12 dichotomous questions is approximately .025 (onetailed). For the image items, 71010 of 3-year-olds, 67% of .Q-year-olds,and 92070 of 5-year-olds achieved this level of performance. For the hidden object items, the corresponding percentages were 9290, 92%, and 9290, and for the visible object items, 96%, 96%, and 100%. At all ages, therefore, a large majority of children were consistently correct in their responses.

c. Explanations. Negative judgments were correct for see, touch, and publicness questions on both mental image and hidden object items, so, as in Experiment 1, any differences in the children’s explanations for these judgments would be especially revealing. Explanations were coded into one of the four following mutually exclusive categories, or into a residual

Understanding of Mental Phenomena

65

category that included “I don’t know,” incomprehensible replies, and refusals to answer. 1. Mental: The child alludes to the mental status of the entity by using a mental term in the explanation (e.g., “It’s just imagination.” “I’m just thinking in my mind.”). 2. Not real: The child says that the entity cannot be seen, touched, or seen by someone else because it is not real. 3. Inaccessible: The child invokes the entity’s inaccessibility to justify a negative response (e.g., “It’s inside my head.” “It’s under the box.’’ “I can’t put my hand through the box.”). 4. Impossible: The child states without further elaboration that he or she cannot perform the action in question (e.g., “Nobody could do that.” ‘‘I couldn’t if I tried.”).

Intercoder reliability (agreements divided by agreements plus disagreements) between two independent coders on approximately 25% of the explanations was .92. The four categories captured 82% of the explanations. Figure 4 displays star graphs for these explanations. For hidden objects, children gave essentially only two types of explanation. In the great majority of cases (8O%), they explained their answers by explicitly invoking the inaccessibility of the object. Though inaccessibility explanations were also the most frequent type on mental image items (43% or 158 explanations), they occurred only about half as often as on hidden object items (80% or 323 explanations): x2(1) = 28.09, p c .001. Moreover, of the inaccessibility explanations for mental images, 59% were of the form “because it’s in my head” or “brain.” This sort of explanation was never given on hidden object items. Explanations in terms of the impossibility of the action were given about equally often on mental image (15%) and hidden object (11%) items: ~ ~ ( =1 )1.22, not significant. Most revealing was the fact that children invoked types of explanations on mental items that they never used for hidden objects: In 15% of their explanations about mental images, they either used a mental term to refer to the mental status of images, or they claimed that images are not real. The number of children giving mental and/or not-real explanations for mental image questions increased with age: 13%, 25%, and 50% for 3-, 4-, and 5-year-olds, ~ ~ ( = 2 )6.00, p < .05.

3. Discussion The overall correspondence of their judgments to the ideal pattern clearly suggests that children of preschool age are aware of having mental images and do not systematically attribute to mental images the properties of physical objects. The deviations from the ideal pattern that did occur generally support

David Estes et al.

66

lmaae

Items

Hidden Obiect Items

Mental

Mental

3-Year-OldS 100%

Not Real

Impossible

4-Year-Olds

T

T l

:0 1

lpo%

l

Not Real

Impossible

5ili+

Not Real

Impossible

t+oyo

Inaccessible

Inaccessible

Fig. 4. Smr graphsfor children'sexplanations on image and hidden object items in Experiment 2. (Similar graphs for Experiment 3 are essentially identical.)

the conclusion that most children were in fact comprehending the questions about mental images and answering appropriately. Inspecting a mental image is a quasi-visual experience and sufficiently like visual perception that the metaphor of seeing is universal (Shepard, 1984). The fact that for older children deviations from the ideal pattern of responses on mental image items were considerably greater on the see question than on the touch and publicness questions suggests that they were undergoing the experience that is the basis for this metaphor. That the 3-year-olds did not display the same pattern is open to various interpretations. One might speculate that the experience of

Understanding of Mental Phenomena

67

forming an image was not as vivid for them as for the older children and thus not as likely to evoke the notion of seeing, or that younger children are more literal or restrictive in their use of the term see. Performance on the transformation questions and probes also suggests that these preschoolers were forming and transforming mental images in response to the instructions and that they understood that thoughts but not physical objects can be manipulated by mental effort alone. C. EXPERIMENT 3: MENTAL IMAGES VERSUS PHOTOGRAPHS

Despite the generally high level of performance in Experiment 2, the youngest children still gave some apparently realistic responses. Three-yearolds and some 4-year-olds at times said that they could touch their mental images and that another person could see their mental images. When this kind of error occurred, their comments often suggested that they were answering about the object they had been asked to imagine (e.g., “Yes, I could touch it because it’s right over there on the floor where you put it”). Occasionally, their explanations of correct judgments about mental images also indicated that they might have been answering about the object itself (e.g., “I can’t touch it because you took it away”). As noted in Experiment 1, such comments may indicate that young children at times interpret questions about mental entities to be about the object represented-answering about referents when asked about representations. This error seems to be of the same general kind as those that Piaget (1929) observed when he questioned children about words. He found that young children will answer yes to questions such as “Is the word needle sharp?” This finding led Piaget to claim that young children believe words have the properties of the things to which they refer, a symptom of the child’s deficient ontology he termed nominal realism. Markman (1976), however, has questioned whether errors of this kind should be taken as evidence that children have such a bizarre notion. According t o Markman, a more plausible interpretation of these findings is that children have difficulty interpreting such questions correctly. In her research, she found that, consistent with Piaget’s results, 7- and 8-year-olds agreed that “the word car has wheels.” When given a choice, however, (e.g., “What has wheels, a car or the word car?”), they consistently answered correctly, demonstrating some knowledge of the distinction between word and referent. Rather than being nominal realists, young children may have a more straightforward and understandable tendency to construe questions about words to be about their referents. Our claim is that this tendency to interpret questions about representations to be about referents is a general one that occurs not only with words, but with other representational entities as well, including mental images. One aim

68

David Estes et al.

of Experiment 3 was to examine this possibility directly, which we did in two ways. First, we supplemented the method of Experiment 2 by adding a choice procedure similar to Markman’s. Second, we directly compared children’s conceptions of mental and physical representations by asking them to make judgments about both a mental image of an object and the object’s photograph in a closed container. An inaccessible physical representation (a photograph in a box) is as close a “close impostor” for a mental image as we could devise. Both are representations, in that they depict something else, and neither affords behavioral or sensory contact. Furthermore, we referred to them in similar terms as “the picture in your head” and “the picture in the box.” If children were subtle ontological realists, then they might well conceive of mental images as inaccessible physical representations, as real “pictures in the head.” Forcing children to make judgments and provide explanations about the differences between images and hidden pictures might reveal misconceptions not revealed in Experiment 2, or it might reveal further sophistication in children’s understanding of these fundamentally different entities. A second extension of Experiment 2 involved the elimination of all mental terms from the procedure. In Experiment 3, no common mental terms were used by the adult at any time during the session. In Experiment 2, the terms “think” and “thinking” were used in the instructions and in the questions about mental images (e.g., “Close your eyes and think about the cup; try to make a picture of the cup in your head”). Children’s performance in our prior studies most likely cannot be accounted for by attributing to them some subtle miscomprehension of ordinary mental terms, one that would allow them to answer our questions appropriately even without a reasonable conception of the mental phenomena to which those terms refer. Still, this interpretation remains a logical possibility. For example, from adults’ use of mental terms (e.g., “You know, I was thinking about that bird we saw in the woods yesterday. I’ve never seen one like it.” “I’m thinking about going to the store. Do you want to come along?”), young children could infer that the objects or events mentioned in statements containing mental terms are often not present. If not present, they cannot be seen, touched, or used. Generally appropriate use and apparent comprehension of mental terms could thus mask a failure to understand that such terms refer to immaterial, subjective mental states or processes. Our previous research argues against this hypothesis, but some version of it could conceivably be true. Piaget in fact argued explicitly that the appearance of mental terms such as think in the speech of the child at around 3 years of age does not mean that “children themselves are conscious of the duality” between thought and things (1929, p. 43). By eliminating use of any mental terms at all in our procedures, we hoped to provide decisive evidence against

Understanding of Mental Phenomena

69

the general claim that correct performance on our tasks could represent some sort of facile but incorrect comprehension of everyday mental terms. 1. Method a. Subjects. A total of 60 preschoolers participated: 12 3-year-olds (mean 3:8; range 3:5 to 3:ll); 24 4-year-olds (mean 4 5 ; range 4:l to 4:ll); and 24 5-year-olds (mean 5:6; range 5:O to 5:Il); 37 were boys. Forty of the participants were enrolled in the same preschool as that used in Experiments 1 and 2; the rest were enrolled in a different school of similar ethnic and socioeconomic make-up. lbenty of the children who participated in this study had also participated in Experiment 2.

b. Design. In this study, two kinds of representational entity were contrasted-a mental image and a photograph in a closed container. Five questions were asked about each of these entities. Four of the questions were the same as those asked in Experiment 2; there was also one question about the possibility of using the representation (either the image or the photograph) for some function for which the actual object could be used, termed the function question in Fig. 5 . The upper left portion of Fig. 5 shows the overall design of the study by depicting the ideal pattern of responses to these questions. As is clear there, mental images and inaccessible photographs are alike in that neither can be used in the same way as the objects they represent, and neither can be seen, touched, or seen by someone else. They differ in that mental images can be mentally transformed but inaccessible photographs cannot. If preschool children understand how mental and physical representations differ, then their judgments should conform to this ideal pattern. They also should tend to use different kinds of explanations on those questions for which the same yes/no judgment is correct for both entities.

c. Procedure. The “happy smile” warm-up task used in Experiment 2 preceded the substantive items, but with the term “think” eliminated. One 3- and two 5-year-olds claimed to be unable to form mental images in the warm-up task and were not tested further. Each child was questioned about both a mental image of and a photograph of two familiar objects (a cup and a pencil), for a total of four items. The instructions used to generate mental images were the same as in Experiment 2, except that mention of thinking was eliminated. Thus, children were simply asked to “Close your eyes and try to make a picture of the (item) in your head.” Similarly, the transformation question was the same as before, but

David Estes et at.

70

PATTERN

image

I function see

t

touch public transform

yes

lOO%t

3-YEAR-OLDS

50v01 nofunction see

image

I

I

I

photo //

touch public transform

/ photo

Fig. 5. Responses to the five questionsfor the two fypesof entity in Experiment 3. The top left figureshows an idealizedpafternof correct responses. The otherfiguresshow actual performance by each age group.

with mention of thinking eliminated (e.g., “Without moving your hands, can you make that picture of the cup in your head turn upside down?”). The functions used for the function questions were, for the cup, “use it to get a drink of water,” and for the pencil, “use it to draw something with.” The transformations for the transformation questions were, for the cup, “make it turn upside down,” and for the pencil, “make it move up and down.” On both mental image and photograph items, the actual object was always present on the table in front of the child during questioning. This procedure makes the mental image items different from those in Experiment 2, in which the actual object was removed before questioning children about the image. This variation means that across the two studies we have questioned children about their images of an object when that object was and also was not present and visible. On photograph items, the function question was always asked before the photograph was placed in the box so that children would not answer negatively simply because the picture was inaccessible. One additional filler question, requiring a yes answer, was included on the photograph items so that children could not always be correct on these items simply by answering no to every question. This question was, “Can you open the box and take it out?”

Understanding of Mental Phenomena

71

Finally, two sorts of probe procedures were used to follow up initial responses. First, as in Experiment 2, children who denied being able to transform a mental image were asked to try to do so. This probe was also used after the transformation question on one image and one photograph item per child, regardless of the initial response. As in Experiment 2, the purpose was to assess whether the probe itself induced children to change their answers. Second, errors on the other four questions (see, touch, publicness, and function) on both mental image and photograph items were followed by a clarification-choice procedure. In this procedure, children in effect were asked to state whether it was the object or its representation they were referring to in their answers. For example, if a child said she could touch the picture in her head, she was asked, “Is it the cup on the table or the picture of the cup in your head that you can touch with your hands?” The same question was then asked about the other possibility as well. If, for example, the child answered the previous question by saying that it was the cup on the table that she could touch, then she would be asked, “Well, what about the picture of the cup in your head, can you touch it with your hands?” This procedure made it possible to determine whether the child attributed the property in question to the representation, the referent, or both. To test the effect of this clarification-choice procedure, it was also used following one image question and one photograph question when the child‘s response was correct. For each kind of item, half the children received this control probe on a see question and half received it on a touch question. Explanations were requested for all answers when feasible. Requests for explanations were omitted when a child had reiterated the same explanation for several answers in a row, provided no explanations for several answers in a row, or began to show signs of fatigue or irritation.

2. Results a. Judgments. Preliminary analyses revealed no effects of gender, object (pencil or cup), preschool, or participation in the first study, and no interaction among these variables. As can be seen in Fig. 5, correspondence to the ideal pattern was generally quite good. Children do not believe that a mental image of an object is simply an inaccessible picture of the object. Thus, despite some deviations from the ideal pattern, the main result is the consistently correct performance at all ages. Mean correct responses, defined as in the ideal pattern, exceeded a chance level of 50% to all questions by all age groups, ts > 2.23, ps < .05, with only one exception: The 3-year-olds’ performance on the photograph transformation question did not differ significantly from chance.

David Estes et al.

72

Further analyses support this graphic depiction of the results. A 3 (Age) x 2 (Entity) x 5 (Question type) repeated-measures analysis of variance on correct responses revealed no main effect of Age or Entity. There was a main effect of question, F(4,228) = 10.47,p < .01, subsumed under Entity x Question, F(4, 228) = 16.50, p < .01, and Question x Age interactions, F(8, 228) = 2.13, p < .05. Conceptually, the Entity x Question interaction is of most interest, and the following comparisons help to elucidate it. Performance on the touch and publicness questions did not differ between image and photograph items, but performance on see, function, and transformation questions did. Performance on the see and function questions was somewhat better on photograph items than on image items, ts(59) > 3.69, ps c .005. Surprisingly, performance on the transformation question was poorer on photograph items than on mental image items, t(59) = 3.97, p < .005. Figure 6 helps to clarify the probable source of children’s errors on the transformation question for the photograph items. The left side of the figure shows performance on this question for each of the two item orders. Children for whom the image items preceded the picture items were incorrect on 48% of the picture transformation questions (i.e., half the time, they claimed to be able to transform the picture in the box). In sharp contrast, children who were questioned about picture items before they were asked any questions about mental images were incorrect on only 7% of their judgments. Performance on the image transformation questions, shown in the right half of Fig. 6, did not differ for the two orders. A likely explanation for this finding is that children who were first asked image transformation questions were

Ye5

100%

-

PHOTO

IMAGE

-

50%

0no

-

;

\,

I

I

Understanding of Mental Phenomena

13

encouraged to erroneously construe picture transformation questions to be about a mental image (of the picture) as well. These analyses are on children’s final answers. They take into account responses to both the probe on the transformation question and to the clarification-choice procedure on the other questions. We also examined the effects of these two probe procedures on children’s answers. Figure 7 compares initial and final performance on each question (i.e., both with and without the transformation probe and clarification-choice procedure). Initial and final responses mirror one another closely, but with some important differences. These differences were always in the direction of greater accuracy. Inspection of Fig. 7 shows eight questions on the two items where children’s responses could have either improved or deteriorated on average in response to the probes; in 8 of 8 instances, mean scores improved (p < .004,binomial). Thus, probing image transformation questions appropriately yielded more yes responses, but probing photograph transformation responses appropriately yielded more no responses. Similarly, on the clarification-choice probe, children typically correctly indicated that they could not touch the image or the hidden photograph, but they could touch the actual object. When these correct answers were probed, they did not change them. The data therefore show that the probe procedures did not simply induce children to change their initial responses; instead, children discriminately corrected initial mistakes.

b. Individual Analyses. Individual performance was consistent with the group data. Each child answered 10 questions about mental images and 10 questions about photographs. The probability of answering correctly by

IMAGE

PHOTO

initial initiol

/

Fig. 7. Performance in Experiment 3 before (initial) and after (final) the transformation question probes and the clarification-choice probes.

74

David Estes et al.

chance 8 or more of 10 dichotomous questions is approximately .05 (onetailed). On the image items, 56 of 60 children reached this level of performance when final answers are taken into account (one 3- and three 5-yearolds failed to reach this criterion). On the picture items, 59 out of 60 got eight or more correct when their final answers were taken into consideration.

c. Explanations. A negative response is correct on four of the five questions for both mental images and photographs (see Fig. 5 ) , so on these questions, the justifications that children gave for their answers provide needed evidence about their understanding of how mental and physical representations differ. Children in this study used the same kinds of explanations that were used in Experiment 2. Analyses of these categories of explanations replicated the essential findings in Experiment 2. In addition, however, one type of explanation occurred in this study that did not occur in Experiment 2. On photograph items, children frequently (16%) justified their negative answers by simply noting that the entity in question was “just a picture.” Only two explanations of this kind (0.6%) were ever used for answers to questions about mental images. The absence of this explanation for the image items is especially revealing because mental images were explicitly termed “pictures in your head” in both Experiment 2 and Experiment 3. The percentage of children using at least one such “just a picture” explanation for the photograph item was 42070, 58%, and 79% for 3-, 4-, and 5-year-olds, respectively. One further aspect of children’s explanations is informative. As in Experiment 2, children explained their judgments about mental images by referring to the mental status of images. Mental explanations for mental images accounted for 10% of the explanations for images in Experiment 3 (and 0070 for pictures) and 13% in Experiment 2 (and 0% for physical objects). In such explanations, mental status is attributed to the entity by the use of an appropriate mental term-for example, “it’s just imagination”; “it’s only a thought”; “I’m just thinking in my mind.” In Experiment 3, because mental terms were never used by the adult in talking to the child, mental explanations represent spontaneous invocation of mental terms and attributes by children. The quantitative data presented in Experiments 2 and 3 d o not do justice to the sophistication of some of the explanations provided by these young children. Table V presents some examples of children’s mental explanations. Across the two experiments, only 4 out of 36 of the 3-, 10 out of 48 of the 4-, and 24 out of 48 of the 5-year-olds gave mental explanations. This is much less use of mental explanations than in Experiment 1, but there are good reasons for this difference. First, Experiment 1 was designed primarily to elicit explanations, so more explanations were requested than in the other two

Understanding of Mental Phenomena

75

TABLE V Sample Explanations from Experiments 2 and 3 Randomly sampled mental explanationsu

Other informative explanationsb

3-Year-olds (Transform image?) Yes, because my head thinks. (Touch image?) No, because it’s my imagination. (Image public?) No, people can’t see my imagination. (Touch image?) No, imaginations is imaginations. (Transform image?) Yes; can’t really do it, but I can pretend to d o it.

3-Year-olds (Image public?) No, but you could see a cup in your head. (Transform image?) Yes, because brains can d o that. (Touch image?) No, because it’s not real.

4-Year-olds (See image?) No, because it’s imaginary. (Image public?) No, ’cause you can only think about it. (Image public?) No, because it’s my imagination. (Transform image?) Yes, because I have dream- hands. (See image?) No, ’cause you’re just thinking about them.

4-Year-olds (Transform image?) Yes, your mind is for moving things and looking at things when there’s not a movie or a TV around. (Image public?) No, because it’s only in my head, but you could make it in your head . (See image?) People can’t see my imagination.

5-Year-olds (Transform image?) Yes, I just could; I could think it right now. (See image?) No, because you just think about it in your mind. (Touch image?) No, not in your head; ’cause you can only think. (Transform image?) Yes, because I would imagine it going up and down. (Image public?) No, ’cause I‘m thinking about them.

5-Year-olds (Transform real scissors?) No, because that doesn’t respond to my head. 1 could think and they wouldn’t open because my head doesn’t connect to that. (Touch image?) How can you reach inside your head? Besides it’s not even there. (Transform image?) I thunk a special way. I can make things come true in my head.

uFive randomly sampled mental explanations. bThree explanations, from three different children, that clearly capture the mental, nonphysical character of thought.

76

David Estes et al.

studies. Further, in Experiment 1, we counted explanations of the type “it’s only in his head” or “brain” as mental explanations. In Experiments 2 and 3, we did not. In Experiments 2 and 3, the experimenter used the phrase “picture in your head” to present the task; therefore, to be conservative, we did not count explanations of this sort as mental. Despite a lesser frequency of mental explanations, however, children’s mental explanations are still informative. Table V presents five randomly sampled mental explanations at each age. This sample gives a flavor for children’s typical mental explanations. In addition, Table V presents three different explanations (from three different children at each age) that seemed to us especially clear or cogent and give an indication of the sophisticated responses that many children at times provided.

3. Discussion The results of Experiment 3 confirm and extend those of Experiment 2. Preschoolers distinguished between mental images and inaccessible physical representations in their judgments and explanations, even though the two were described to them in similar terms. The results indicate a basically solid understanding of the distinction between mental and physical representations. This high level of performance was achieved even though no conventional mental terms were used by the adult at any time during the procedure. This contradicts the possibility that preschoolers’ knowledge of mental phenomena might be limited to some deceptively useful miscomprehension of common mental terms. The results of Experiments 2 and 3 fit together nicely. Children in these studies consistently judged that mental images cannot be seen, touched, seen by someone else, or used in the way a real object can be used, but they can be mentally transformed. Their judgments about mental images differed appropriately from their judgments about inaccessible physical representations and about actual objects. The overall pattern of their judgments for these different kinds of entities conformed closely to an ideal pattern of judgments. Importantly, errors that might be taken as evidence of Piagetian realism in both experiments often disappeared in Experiment 3, when children were allowed to choose whether it was the mental image or the object itself that had the property in question. Equally important, this clarification-choice procedure improved performance on photograph questions, suggesting that when they do occur, errors on questions about mental images specifically, and about representations generally, are often due to misinterpretation of questions about representations as being questions about their referents instead. Such errors therefore do not seem to index a lingering realistic misconception about the nature of mental entities. Errors that remained after the clarification-choice and transformation probe

Understanding of Mental Phenomena

77

were informative. In both experiments, children at times claimed they could see a mental image. The clarification-choice procedure did not reduce errors on the see question for mental images to the extent that it did on other questions (see Fig. 7). This suggests that these apparent errors on image see questions were due not to a misinterpretation of the question, but to the quasi-visual nature of the experience. This is the pattern that would be expected if children were actually forming mental images in response to the instructions. Note in this regard that children tended to say that they could see their images, but others could not, confirming that they recognize the private nature of their mental experiences. In fact, summed over Experiments 2 and 3, the children provided 77 instances in which they differed on their answers to the see and publicness questions. In 60 of these instances (780ro), they said they could see their image but someone else could not. These children also often gave cogent, adult-like explanations for their correct responses. In explaining why a mental image cannot be used like the object it represents, for example, older children often spontaneously invoked its status as a mental entity by using mental terms in their explanations. They never used such terms for physical entities. Conversely, in explaining why a photograph cannot be used for a typical function of the object it represents, they often referred to its status as a picture. They almost never claimed that mental images were “just pictures,” even though both mental images and photographs were described by the experimenters as pictures. Children’s pattern of judgments, coupled with their explanations, provide decisive evidence that young children have appropriately mentalistic conceptions of mental images. Our use of children’s explanations as evidence for their understanding of mental phenomena, and especially our corroborative use of specific examples, deserves further comment. It could have been the case that preschool children can respond correctly to the kinds of simple judgments we asked them to make about how mental and physical phenomena differ, yet still be incapable of providing reasonable explanations for their answers. Had this occurred, it would suggest that, although young children have considerable implicit knowledge of mental phenomena, they might lack anything like the kind of explicit knowledge that adults have. That a substantial number of children in these three studies provided sensible and often elaborate explanations indicates that preschoolers frequently have an explicit, articulate understanding of mental phenomena. D. COMPARISON WITH LAURENDEAU A N D PINARD

We wish to be clear that our data in these three experiments do not represent a simple failure to replicate Piaget’s (1929) prior findings or those of

78

David Estes et al.

other investigators such as Laurendeau and Pinard (1962). We believe that these earlier findings are replicable but misinterpreted. The apparent differences between our findings and earlier findings could occur for several reasons. Other investigators might simply be more sensitive elicitors of children’s true beliefs than we are. Alternatively, we might have tested an overprivileged sample of children who have arrived at their sophisticated ideas atypically early in development. The difference might also reflect important cohort differences; our children were tested in the mid-1980s’ Laurendeau and Pinard’s in the late 1950s, and Piaget’s in the 1920s. Perhaps with increased early childhood education, or increased exposure to television and computers, the young children of today are simply more knowledgeable about these matters than children in prior years. All these possibilities can be refuted by showing that the children we tested and the children tested by Piaget or by Laurendeau and Pinard are in fact giving similar responses. Our test of this possibility involved taking the explanations of children in some of our studies and classifying them according to Laurendeau and Pinard’s coding system. If our subjects have different conceptions from theirs, then our children and theirs should still look substantially different in their responses. If, however, the two sets of data look quite similar when children’s responses are evaluated with a comparable method, then the interpretation of those responses is what is at issue. We used Laurendeau and Pinard’s (1962) system of analyzing children’s responses because they spelled out their coding system more precisely and in more detail than did Piaget (1929), and their results replicated Piaget’s. We begin with a brief overview of their study. Ldurendeau and Pinard (1962) interviewed children about only one sort of mental phenomenon-dreams. Here are some of their questions: “Where does a dream come from?” “Where are dreams made, where do they come from?. . .Do they come from inside of you, o r outside of you?” “Who makes the dream come?. . . Is it you or someone else?” “When you dream that you are playing in the street, where is your dream? In the street, or in your room?” “When your mother is in your room, can she also see your dream? And I, if I were in your room, could I see your dream?” “When, during the night, you dream that you are playing, are you really playing?. , .Is it the same as when you play for real?. . .Then, are our dreams true?” (pp. 63-65)

From our perspective, these kinds of questions are objectionable in several ways: Some are unclear; others seem to suggest that dreams are substantial entities that are made of something and come from somewhere. But our primary focus for the current analysis concerns Laurendeau and Pinard’s

Understanding of Mental Phenomena

19

system for categorizing and interpreting children’s responses. What they did, following Piaget, was to evaluate the overall quality of children’s responses to all questions by classifying each child’s pattern of responses into one of four categories: 1. Incomprehension or refusal. The child provides no comprehensible answers. This is evident in answers such as, ”I don’t know” or “Because.” 2. Integral realism. The child describes dreams in terms suitable for external concrete physical objects. For example, “The dream is in the room,” “on the wall,” “in front of me” (p. 107). Dreams can be seen by others, thus, “Mother can see my dream,” or “Mother can’t see it,” because the room is too dark (p. 108). Dreams are made of physical substances like “cloth,” “wood,” “skin” (p. 109). And dreams can be touched (p. 110). 3. Mitigated realism. The hallmark of mitigated realism is that at times the child gives integral realism answers but at other times gives subjectivist answers. In addition, dreams may be described as internal or even as intangible but in ways one would use to describe internal or intangible physical entities, such as internal organs or moving pictures. Thus, the dream is “in the heart” (p. 114) or “in the tummy,” “in the eyes,” (p. 113). Children might describe dreams as in the head, but still, “they will uphold the possibility that others can see the dream inside the head” (p. 114). Or, if dreams are described as events or even as pictures they take place “outside” (p. 116) or “in front of ourself’ (p. 117) and “the dreamer is not the only one who can look at the spectacle” (p. 119). 4. Subjectivism. Dreams are described as subjective, private, mental experiences. For example, it’s “in the mind,” “it’s in the imagination,” (p. 121), it’s “inside of me, in my thinking” (p. 125). Laurendeau and Pinard’s data are shown at the top of Table VI. Incomprehension was fairly common, especially at the younger ages. Excluding incomprehensible replies, in the preschool years, the responses were split between the two sorts of realism, with mitigated realism predominating as children get older. Subjectivism appeared only late in this age range, at about 6 years. In order to compare our data with theirs, an independent rater, uninformed about our hypotheses, coded children’s explanations in our studies according to Laurendeau and Pinard’s coding system. All the explanations collected about mental entities in Experiment 1 (nine explanations per child) and the explanations about mental entities in Study 2 of Wellman and Estes (1986) (eight explanations per child) were individually coded as representing one of the four preceding categories. Then each child was assigned to one of these four categories on the basis of his or her overall responses. Each child was assigned to the category of the majority of his or her explanations, except that if a child showed a mixture of answers in Categories 2 and 4, 3 and 4,

David Estes et al.

80

TABLE VI Percentages of Children in Each Laurendeau and Pinard Category ~~~~~~

Age

N

6 5: 5 4: 4 3i

49

6

31

~~

Incomprehension

Integral realism

~

Mitigated realism

Subjectivism

Laurendeau & Pinard (1962)

5;

5 4: 4 3:

46 50 49 -

36 36

10 24 46 55

-

10

38

-

-

24 26 20 -

38 28 24

-

Experiment 1 and Wellman & Estes (1986) 0 6 16

8

53

-

-

22 39

55 29

41 13 0 0

39 17 16

or 2, 3, and 4, he or she was assigned to the category of mitigated realism. This procedure is in accord with Laurendeau and Pinard’s description of mitigated realists as evidencing some subjectivist answers coupled with other realist responses. Thus, in our procedure, three fourths or more of a child‘s responses (6 of 8 or 7 of 9) had to be subjectivist before a child was considered as evidencing subjectivism. When one of the authors coded the responses of 55 children, agreement between him and the independent coder was .89. With this coding system, we obtained the data presented in the bottom portion of Table VI. Some differences are apparent across the two data sets. For example, we got much less incomprehension and refusal to reply in our data, probably because our questions are more straightforward and less confusing. But excluding incomprehension, our children’s responses were very similar to Laurendeau and Pinard’s. Our data, as theirs, yields essentially a split between the two sorts of realism in the younger children, with mitigated realism predominating as children get older. Only in the oldest children do we find much subjectivism. To confirm these conclusions, we compared the Laurendeau and Pinard 6-year-olds with our 54-year-olds, their d-and 5-year-olds with our 4i-year-olds, and their 4-year-olds with our 3Z-year3 olds. The distributions of children across the three substantive categories (integral realism, mitigated realism, and subjectivism) did not differ in any of these comparisons: x’s (2) < 5.65, p > .05.

Understanding of Mental Phenomena

81

Our conclusion from these closely comparable findings is that the differences between our estimation of children’s knowledge of mental phenomena and Laurendeau and Pinard’s cannot be simply a failure to replicate. Instead, we argue that the interpretations arrived at by Laurendeau and Pinard and by Piaget from such data are flawed for two reasons. First, interview studies such as Piaget’s and Laurendeau and Pinard’s include only one half of the needed experimental design. In general, if we wish to determine whether someone distinguishes or confuses two sorts of things, then we need to compare his or her responses to both things. For this reason, we have included in our studies a variety of informative contrasts. Children’s answers to questions about mental entities can thus be directly compared to their answers about various kinds of physical entities. Our use of this procedure clearly revealed that children can make the appropriate, at times quite subtle, distinctions. The second related flaw in Piaget’s and Laurendeau and Pinard’s method is that in the absence of directly comparable responses about both mental and physical entities, these investigators were encouraged to interpret young children’s explanations of mental entities literally. Young children’s statements that “dreams are smoke” are coded as meaning dreams are actually smoke rather than dreams are like smoke. We claim that such literal interpretations are incorrect. Children’s statements should instead be seen as attempts to point out some informative analogies and similarities, not as literal attempts to specify identities. For two reasons, such statements on the part of children should be taken as nonliteral but informative attempts to talk about difficult topics. First and more decisively, our complete data directly show that young children are aware of the appropriate differences. They know that mental entities are not literally smoke, shadows, physical objects, pictures, or even inaccessible pictures or objects. The second reason is that in speaking nonliterally about the mind, young children are only engaging in common adult practice for talking about mental phenomena. Adults say that dreams are like pictures, and we “see” our images. We use the language of the external world nonliterally to refer to the mental world. That we do so does not mean we are ontological realists. The same seems to be true of young children. It is important to note that given either our methods or Laurendeau and Pinard’s, young children do at times say they can touch or see their dreams or thoughts. Such errors, however, seem best explained as revealing a tendency of young children to interpret questions about mental representations (a d m m about the ball) as being questions about their referents (the ball). These responses, although interesting, also cannot be taken as evidence of ontological realism because young children make similar errors about concrete external representations, such as photographs, and they can correct such errors when asked to do so.

82

David Estes et al.

VI. General Discussion Preschool children in these studies consistently judge that mental phenomena-thoughts, memories, dreams, and mental images-differ in fundamental ways from physical phenomena, even from potentially confusing “close impostors.” These preschool children also frequently gave cogent and appropriate explanations for their correct responses. The existence of a substantial number of explanations of this kind, occurring in an appropriate context, is by itself strong evidence that young children understand the distinction between mental and physical phenomena. In combination with their consistently correct judgments, the evidence is decisive. This research further demonstrates that preschool children can apply their understanding of the mental-physical distinction to themselves and to others, can take their own mental experiences as objects of reflection and discourse, have at least a rudimentary capacity to transform mental images, and understand the distinction between a mental representation and what it represents. These understandings and capacities were as apparent in 3-year-old children as in 5-year-olds. Furthermore, this early understanding of mental phenomena was not confined to some facile but superficial interpretation of common mental terms. Our findings thus warrant some strong conclusions. The primary conclusion is that children’s understanding differs diametrically from the traditional Piagetian view: Let us imagine a being, knowing nothing of the distinction between mind and body. Such a being would be aware of his desires and feelings but his notions of self would undoubtedly be much less clear than ours. Compared with us he would experience much less the sensation of the thinking self within him, the feeling of being independent of the external world. The knowledge that we are thinking of things severs us in fact from the actual things. But, above all, the physical perceptions of such a being would be entirely different from our own. . . . We shall try to prove that such is the case with the child. The child knows nothing of the nature of thought, even at the stage when he is being influenced by adult talk concerning “mind,” “brain,” “intelligence.” (1929, p. 37)

Piaget was referring very generally here to the mutual implications of the child’s sense of self, the child’s understanding of mind, and the child’s conscious phenomenology. But if this description or anything like it were true, how could the children in our studies have performed as they did, or even comprehended our tasks at all? According to the traditional Piagetian view, it is as if young children live in a different world, one in which the distinction between the mental and the physical is nonexistent. In sharp contrast to this view, preschool children in our studies respond as though they and the experimenter were inhabiting the same world and dividing it into the mental and the physical realms along essentially the same boundary.

Understanding of Mental Phenomena

83

These findings shed light on other recent research on children’s understanding of mental phenomena, their “theories of mind” as this topic has become known. From our research, it is clear that even very young children are ontological dualists rather than ontological realists as we have defined these terms. This conclusion still leaves open the nature of children’s epistemology, their understanding of the acquisition, transformation, and origins of knowledge. Recent research on this question shows that younger preschool children fail to understand how knowledge is derived from perceptual and communicative sources (Wimmer, Hogrefe, & Sodian, 1988), fail to understand that the same external information can mean different things to different people (Chandler & Helm, 1984; Taylor, 1988), fail to recognize that appearances and reality do not always straightforwardly correspond (Flavell, Flavell, & Green, 1983), and fail to conceive of the mind as an informationprocessing, interpreting, construing device (Johnson & Wellman, 1982). Substantial development on these topics occurs by middle childhood. These changes have been characterized as constituting a progression from a copy to a constructivist theory of knowledge (Chandler & Boyes, 1982), an encounter to an information-processing epistemology (Wellman, 1988), or a passive (Level 1) to an active (Level 2) theory of mind (Flavell, 1988). Our findings, taken in conjunction with this other research, indicate that children’s first theory of mind is characterized by ontological dualism, albeit probably coupled with epistemological realism. Young children understand the fundamental distinction between thought and objects, even while assuming that external objects directly generate thoughts, which always objectively mirror reality. This revised description, underwritten by our own studies and other related research, is an important advance. It helps to explain young children’s heretofore puzzling mix of competence and incompetence, their uneven understanding of different aspects of mental phenomena. It also helps to elucidate what would otherwise be inexplicable: If young children were really incapable of distinguishing between the mental and physical realms, how could we ever communicate with them as well as we do? None of the previous research on children’s theories of mind directly addressed the topic of children’s introspection. Conceptual knowledge about the mind does not necessarily mean that children can contemplate and discuss their own mental experiences. Forming and transforming a mental image, however, certainly counts as a mental experience. Our data thus suggest that young children can reflect on, report on, and discuss their mental experiences. The pattern of their responses and the appropriateness of their explanations together are convincing. For example, many children claim that they could see their images (or see them “in my head”) but that no one else could see their images, and they claimed that they could mentally transform their images but not physical objects. In sum, children’s reports were sensible and often

84

David Estes et a/.

quite elaborate, similar in fact to what would be expected from adults questioned in the same way, and very different from Piaget’s examples of children reporting on mental arithmetic. Our data do not demonstrate that young children have a fully developed adult-like ability to introspect. Indeed, some have argued that adult introspection is in large part a conventional system of discourse (Lyons, 1986) based on our theories about cognition (Nisbett & Wilson, 1977). The nature of introspection is of course a topic that has engendered a long history of controversy in philosophy and psychology. One issue in this debate concerns whether introspection of the sort we are familiar with as adults first appears early or late in childhood (cf. Lyons, 1986, p. 97). If children take a long time to develop any concept of what is meant by introspection or how to monitor and report their own thoughts, then conceivably introspection is largely a product of learning and socialization. It is notable in this regard that children were easily able to adopt and utilize the “picture in the head” metaphor to refer to their images. They did so even in Experiment 3 in which the experimenter used no common mental terms, the familiarity of which might have provided cues about how to respond appropriately. To verify our intuition that “picture in the head” is not likely to be a familiar phrase to most children, we conducted a computerized search through six of the longitudinal language corpora from the CHILDES language data archive (MacWhinney & Snow, 1985). We found no use of this phrase, or any variations of it, either in the speech of children or in adult’s speech to children. Our experimental findings, showing a rudimentary ability to report and sensibly discuss mental experiences at a very young age, in conjunction with these analyses of everyday speech, suggest that young children can refer to and discuss their own mental states using novel terminology, and before they have had much opportunity to learn the conventions and language of introspection. Our results do not support any particular position on the ultimate nature of introspection, nor do they provide evidence about the origins of this early introspective capacity. But our findings do provide evidence against the claim that there is little basis in untutored experience to inspire introspective reports and that the ability to provide such reports must be slowly acquired through socialization. A capacity for introspection thus seems to be one source of young children’s understanding of mental phenomena and their developing theories of mind (see also Johnson, 1988). The performance of preschool children in our imagery studies demonstrates that a rudimentary ability to reflect on and manipulate one variety of mental phenomena is present very early. Such a demonstration seems inconsistent with a widely held view in the metacognition literature that “the ability to step back and consider one’s own thought as itself an object of thought’’ is

Understanding of Mental Phenomena

85

an acquisition of late childhood or even adolescence (Brown etal., 1983, p. 122). We speculate that preschoolers’ ability to reflect on and transform mental images might be an early manifestation of a broadly applicable and highly valuable capacity to consider objectively some mental entity or process (e.g., an idea, belief, emotion, or intention), to view it not as something static, given, or immutable, but rather as something one can actively examine, manipulate, “play with,” and perhaps attempt to modify. The results of the two studies on preschool children’s understanding of mental imagery also adds to the meager knowledge that exists about the development of imagery. Although the construct of mental imagery has played a central role in psychological theory and especially in theories of cognitive development (Bruner, Olver, & Greenfield, 1966; Piaget & Inhelder, 1971), we still know remarkably little about the developmental course of mental imagery abilities. For example, conflicting claims have been advanced about when children become capable of transformational or kinetic imagery. Piaget and Inhelder (1971) claimed that transformational imagery is acquired only with concrete operations in middle childhood. More recently, some researchers have adapted the chronometric methods used in adult imagery research (e.g., Shephard & Cooper, 1982) to investigate young children’s transformational imagery ability, but this approach has produced equivocal findings. Some studies have yielded evidence for transformational imagery in children as young as 4 years of age (Marmor 1975), but others have failed to replicate these results (Kerr, Corbitt, & Jurkovic, 1980). Our data, though not chronometric, provide strong evidence that children as young as 3 years of age possess some degree of transformational imagery ability.

VII. Conclusion In summary, the studies reported here provide a mix of findings, some of which we believe are definitive and others intriguingly suggestive. That young children are ontological realists can no longer be considered tenable. Our data make it abundantly clear that the beginning theory of mind is one of ontological dualism, which is probably combined with some form of epistemological realism. These conclusions about early understanding of mental phenomena are based on findings concerning young children’s reflection on and conception of their own and others’ mental experience, their awareness, understanding, and control of imagery, as well as their conceptions of other kinds of things, such as photographs, smoke, and shadows. This interwoven set of findings seems fitting-mental phenomena themselves are complex, multifaceted and interrelated. Philosophy of mind and cognitive science take as their domain a host of related phenomena and processes. Young

86

David Estes et al.

children’s conceptions seem to mirror this complexity in scope, if not in sophistication; even very young children are beginning to recognize and think about a variety of mental phenomena. Furthermore, young children demonstrate at least one firm foothold in this complex and potentially confusing domain - namely, a solid and articulate understanding of the fundamental differences between mental entities and physical objects. ACKNOWLEDGMENTS This research was supported by grants from NICHD (HD-00525, HD-22149) and from the University of Michigan (under a Biomedical Research Support Grant) to Wellman. Woolley was supported by a traineeship from NICHD and Estes by a fellowship from the Graduate School of the University of Michigan for part of their efforts on this research.

REFERENCES Anderson, E. (1960). A semi-graphical method for the analysis of complex problems. Technometrics, 2, 387-392. Broughton, J. (1978). Development of concepts of self, mind, reality, and knowledge. In W. Damon (Ed.), New directionsfor child development (pp. 75-100). San Francisco, CA: Jossey-Bass. Brown, A. L., Bransford, J. D., Ferrara, R. A., & Campione, J. C. (1983). Learning, remembering, and understanding. In P. Mussen, J. Flavell, & E. Markman (Eds.), Handbook of child psychology: Vol. 3. Cognitive development (pp. 77-166). New York: Wiley. Bruner, J. S., Olver, R. R., & Greenfield, P. M. (1966). Studies in cognitive growth. New York: Wiley. Chandler, M. J., & Boyes, M. (1982). Social-cognitive development. In B. Ludman (Ed.), Handbook of developmental psychology (pp. 287-402). Englewood Cliffs, NJ: Prentice-Hall. Chandler, M. J., & Helm, D. (1984). Developmental changes in contribution of shared experience to social role taking competence. International Journal of Behavioral Development, 1, 145-156. Dennett, D. C. (1979). Brainstorms. Sussex: Harvester Press. Ericsson, K. A., & Simon, H. A. (1980). Verbal reports as data. PsychologicalReview,87,215-251. FlaveU, J. H. (1979). Metacognition and cognitive monitoring: A new area of psychologicalinquiry. American Psychologist, 34, 906-911. Flavell, J. H. (1988). The development of children’s knowledge about the mind: From cognitive connections to mental representations. In J. W. Astington, P. L. Harris, & D. R. Oldson (Eds.), Developing theories of mind. New York: Cambridge University Press. Flavell, J. H., Flavell, E. R., & Green, F. L. (1983). Development of the appearance-reality distinction. Cognitive Psychology, 15, 95-120. Flavell, J. H., & Wellman, H. M. (1977). Metamemory. In R. Kail & J. Hagen (Eds.), Perspectives on the development of memory and cognition. Hillsdale, NJ: Erlbaum. Johnson, C. N. (1988). Theory of mind and the structure of conscious experience. In J. W. Astington, P. L. Harris, & P. R. Olson (Eds.), Developing theories of mind. New York: Cambridge University Press. Johnson, C. N., & Wellman, H. M. (1982). Children’s developing conceptions of the mind and brain. Child Development, 53, 222-234. Keil, F. C. (1979). Semantic and conceptualdevefopmenf.Cambridge, MA: Harvard University Press.

Understanding of Mental Phenomena

87

Kerr, N. H., Corbitt, R., & Jurkovic, G. J. (1980). Mental rotation: Is it stage related? Journal of Mental Imagery, 4, 49-56. Laurendeau, M., & Pinard, A. (1962). Causal thinking in the child. New York: International Universities Press. Lyons, W. (1986). The disappearance of introspection. Cambridge, M A Bradford Books/MIT Press. MacWhinney, B., & Snow, C. (1985). The child language data exchange system. Journal ofchild Language, 12, 271-296. Markman, E. M. (1976). Children’s difficulty with word-referent differentiation. Child Development, 47, 142-149. Marmor, G. S. (1975). Development of kinetic images: When does the child first represent movement in mental images? Cognitive Psychology, 7, 548-559. Misciones, J. L., Marvin, R. S., O’Brien, R. G., & Greenburg, M. T. (1978). A developmental study of preschool children’s understanding of the words “know” and “guess.” Child Development, 49, 1107-1113. Nisbett, R. E., & Wilson, D. T. (1977). Telling more than we can know: Verbal reports o n mental processes. Psychological Review, 84, 231-279. Piaget, J. (1928). Judgment and reasoning in the child. New York: Harcourt & Brace. Piaget, J. (1929). The child‘s conception of the world. London: Routledge & Kegan Paul. Piaget, J., & Inhelder, B. (1971). Mental imagery in the child. London: Routledge & Kegan Paul. Shantz, C. U. (1983). Social cognition. In P. Mussen, J. Flavell, & E. Markman (Eds.), Handbook of childpsychology: Vol. 3. Cognitive development (pp. 495-555). New York: Wiley. Shepard, R. N. (1984). Ecological constraints on internal representation: Resonant kinematics of perceiving, imagining, thinking and dreaming. Psychological Review, 91, 417-447. Shepard, R. N., & Cooper, L. (1982). Mental images and their tramformations. Cambridge, MA: MIT Press. Thylor, M. (1988). Conceptual perspective taking: Children’s ability to distinguish what they know from what they see. Child Development, 59, 103-718. Wellman, H. M. (1985). The origins of metacognition. In D. L. Forrest-Presley, G. E. MacKinnon, & T. G. Waller (Eds.), Melacognition, cognition, and human performance @p. 1-31). Orlando, FL: Academic Press. Wellman, H. M. (1988). First steps in the child’s theorizing about the mind. In J. W. Astington, P. L. Harris, & D. R. Olson (Eds.), Developing theories of mind. New York: Cambridge University Press. Wellman, H. M. (in press). The early development of memory strategies. In F. Weinert & M. Perlrnutter (Eds.), Memory development: Universal changes and individual differences. Hillsdale, NJ: Erlbaum. Wellman, H. M., & Estes, D. (1986). Early understanding of mental entities: A reexamination of childhood realism. Child Development, 51, 910-923. Wimmer, H., Hogrefe, J., & Sodian, B. (1988). A second stage in children’s conception of mental life: Understanding sources of information. In J. W. Astington, P. L. Harris, & D. R. Olson (Eds.), Developing lheories of mind. New York: Cambridge University Press.

This Page Intentionally Left Blank

SOCIAL INFLUENCES ON CHILDREN’S COGNITION: STATE OF THE ART AND FUTURE DIRECTIONS

Margarita A zmitia DEPARTMENT OF PSYCHOLOGY FLORIDA INTERNATIONAL UNIVERSITY MIAMI. FLORIDA 33199

Marion Brlmutter DEPARTMENT OF PSYCHOLOGY AND INSTITUTE OF GERONTOLOGY UNIVERSITY OF MICHIGAN ANN ARBOR, MICHIGAN 48109

1. INTRODUCTION

11. THEORETICAL OVERVIEW A. SYMBOLIC INTERACTIONIST PERSPECTIVE B. SOCIOHISTORICAL PERSPECTIVE C. STRUCTURAL PERSPECTIVE D. SOClAL LEARNING PERSPECTIVE E. CONTEMPORARY PERSPECTIVES F. SUMMARY 111. EVIDENCE FOR THE IMPACT OF SOCIAL AGENTS

A. B. C. D. E.

RELATIONSHIPS AND COGNITIVE DEVELOPMENT PARENT-CHILD INTERACTION PEER INTERACTION SIBLING INTERACTION SUMMARY

IV. A FRAMEWORK FOR CONSIDERING DEVELOPMENTAL CHANGE IN SOCIAL INFLUENCES ON COGNITION A. BENEFITS OF FRAMEWORK B. DESCRIPTION OF FRAMEWORK C. SUMMARY V. ADDITIONAL THEORETICAL AND METHODOLOGICAL ISSUES A. THEORETICAL ISSUES B. METHODOLOGICAL ISSUES

VI. CONCLUSIONS REFERENCES

89 ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR, VOL. 21

Copyright 0 1989 by Academic Press, Inc. All rights of reproduction in any form reserved.

90

Margarita Azmitia and Marion Perlmutter

I.

Introduction

Since the late 1970s, we have we have become increasingly aware that learning and problem solving are often social endeavors. Variations in social environments have been used to explain differences in individuals’ problemsolving strategies (e.g., Laboratory of Comparative Human Cognition, 1983; Laosa, 1982; Luria, 1976; Wood, 1980), academic achievement (Doise & Mugny, 1984), literacy (McNamee, McLane, Cooper, & Kervin, 1984), numerical skills (Saxe, Geahart, & Guberman, 1984), memorial competence (e.g., Ellis & Rogoff, 1982. Ratner, 1984), and language development (C. P. Jones & Adamson, 1985; K. Nelson, 1981). Social contexts can facilitate learning because partners often provide new information, define a problem in such a way that it becomes manageable, and generate a discussion that culminates in the selection of the best strategy and solution. Kuhn and Phelps (1979, 1982), for example, have suggested that the major obstacle for development may be the abandonment of old, inefficient strategies, not the acquisition of new, more efficient ones. It is difficult to abandon old strategies because they are partially successful and, in a sense, some strategy is better than no strategy. Interpersonal contexts allow individuals to see their inefficient strategy simultaneously in operation with a potentially more efficient strategy, and they can also provide a rationale for why one strategy is better than another, and thus can be catalytic to cognitive growth. Much research (see Paris & Lindauer, 1982, for a review) has shown that children will continue to use a newly acquired strategy only if they are given a rationale for why it facilitates performance. Social interaction does not contribute only to the cognitive aspect of problem solving. A partner can increase one’s task motivation by energizing behavior, making the task appear manageable, providing emotional support in a difficult situation, or making a task more enjoyable. In addition, individuals may be willing to try new techniques because they believe that if something goes wrong, their partner may know how to correct the error; if this is not the case, surely between the two of them they will figure it out. Still, many unanswered questions remain. Most notably, we lack information about how some of the proposed mechanisms influence cognition, that is, what features of interaction promote cognitive growth. In addition, we lack appreciation of the possible ways that developmental level moderates social influence on cognition. In this article, we review research on the relationship between social interaction and cognitive development. We also propose a framework that allows us to explain a wide range of findings and to begin to hypothesize about the nature of changes in social influence across development. Although we argue that social interaction may facilitate cognitive development, we also point to

Social Influences on Cognition

91

some limitations of this facilitation. For example, we suggest that the participants’ cognitive and social skills influence their ability to collaborate and the likelihood that they will learn through collaboration. Also, we acknowledge important individual differences in participation and in the effects of social influence. For example, some individuals may not feel comfortable in a group situation and thus may fail to offer suggestions or seek guidance. Finally, we suggest that some tasks are more amenable to collaborative learning than others. We feel that by examining how social interaction facilitates and hinders cognitive change we will be able to integrate current research and to formulate better research questions. Our major goal is to incorporate the available data into a developmental framework for studying and evaluating social influences on cognition. We accomplish this goal in the following manner. First, we present a brief theoretical overview of approaches to the study of social influences on children’s cognition. Second, we review and integrate research on this topic. Third, we offer a preliminary framework for conceptualizing social influences across development. Fourth, we discuss some additional issues that need to be considered when studying social influences on cognition. Finally, we offer conclusions about the state of the field and some directions for future research. We believe that a developmental approach can add much to our understanding of social influences on cognition. We do not mean to imply that researchers studying this problem have completely ignored developmental issues. Issues of change and growth have motivated most of the investigations of social influences on early cognition. However, we feel that these issues should be made more explicit. For example, some of the generalizations made about social influences on cognition may need to be qualified to accommodate developmental constraints on interaction. Also, social influences on cognition may aggregate over time, and their contribution may change across age. At present, these issues cannot be addressed because, although much attention has been paid to the way in which the social agent’s behavior influences a child’s cognitive performance, comparatively little attention has been paid to the question of how the child’s cognitive and social skills mediate this influence. Moreover, because most studies have not included multiple age groups, little is known about developmental patterns in social interaction and how these patterns influence the type and magnitude of cognitive changes that are observed. Although social influences on cognition have been considered at both a molar level (e.g., influences that arise from individuals’ participation in social institutions such as schools and social classes) and a molecular level (e.g., influences that arise from individuals’ interactions with other individuals), the article is focused only on the latter. For discussions of the former, see Bee, Van Egeren, Streissguth, Nyman, and Fleckie (1969), Hess and Shipman

Margarita Azmitia and Marion IWmutter

92

(1967), Laosa (1980, 1982), and Stevenson (1982). In addition, although the discussion is focused on interactions involving children, this focus does not imply that social interaction does not influence cognition later in the life span. Rather, this focus reflects the preponderance of research that is presently available and our belief that patterns of adult interaction originate in childhood experiences. Thus, to understand later development, we must first understand childhood patterns. Finally, because the emphasis is on conceptual integration, the review of current research is brief and relatively selective. Interested readers are referred to excellent reviews by Ashworth (1979), Doise and Mugny (1984), Laboratory of Comparative Human Cognition (1983), Perret-Clermont (1980), Rogoff, Gauvain, and Ellis (1984), and Wertsch (1985).

11. Theoretical Overview In this section, we review the leading perspectives on social influences on cognition and the factors that contributed t o the growing interest in this problem. We also discuss how each of these perspectives addresses or fails to address developmental issues and assess their current status in the field. A.

1.

SYMBOLIC INTERACTIONISTPERSPECTIVE

Key Assumptions

Models highlighting the contribution of social interaction to development have been available at least since the early part of the 20th century. For example, in his seminal work, Mind, Seg and Society, George Herbert Mead (1934) proposed that cognition originates in social interaction, and that even before infants are capable of symbolic functioning, they engage in a “conversation of gestures” with their caretakers that lays the foundation for cognition. These early nonverbal interactions initially serve only as an approximation of communication because they are maintained exclusively by the caretaker, who behaves as if the infant’s signals have communicative intent (Bruner, 1977). Only when infants begin to anticipate their caretaker’s signals-that is, become capable of reciprocity-will these “conversations of gestures” qualify as true communication. Like linguistic interactions, these early “conversations of gestures” use behaviors that have the same meaning for all members of society; unlike linguistic interactions, these behaviors d o not presuppose any thought in the infant (Perret-Clermont & Brossard, 1985). Mead argued that parents are crucial agents of development because they initially are responsible for initiating and maintaining the “conversations of gestures.” These “conversations of gestures” are subsequently replaced by

Social Influences on Cognition

93

language, which then becomes the primary mechanism through which social interaction creates the mind. For Mead and others in the dialectical tradition (e.g., Riegel, 1979; Vygotsky, 1978), the meaning attributed to symbols (e.g., language) derives from a social process of negotiation among the participants in an interaction. He proposed that individuals form mental representations about others’ behavior that allow them to anticipate their behavior during interaction and thus facilitate the negotiation process. This ability to represent the world mentally frees the child from the here and now and allows him or her to reflect on the past, to participate in the present, and to prepare for the future.

2. Developmental Implications Although Mead’s model has been influential in social psychology, it had not been widely explored by developmentalists. Only recently have developmental psychologists (e.g., Condon & Sander, 1974; Jaffe, Stern, & Perry, 1973; Stern, 1974) attempted to document the parallels between “conversations of gestures” and subsequent linguistic interactions between parents and their infants. Developmentalists’ lack of interest in Mead’s model may have been due, in part, to Mead’s lack of developmental focus. Although he argued for the importance of social influences on cognition throughout the life span, he did not specify developmental sequences. Thus, researchers have had difficulty in generating from his model hypotheses about the process through which children become capable of social negotiation and about how age differences in this ability mediate the nature of cognitive changes. B. SOCIOHISTORICAL PERSPECTIVE

1. Key Assumptions

Vygotsky (1929, 1962, 1978), whose thinking was heavily influenced by Marx’s and Hegel’s dialectical materialism, shared Mead’s view that cognition originates in social interaction and that language plays a crucial role in this process. He believed that development is characterized by the child‘s internalization or appropriation of cultural tools, goals, and activities in order to become a functional member of society (e.g., achieve intersubjectivity). He proposed that one of the impetuses for development is the dialectic between the individual’s maturational growth and personal experience and his or her culture’s tools and activity patterns (Wertsch, Minick, & Ams, 1985). This dialectic motivates the individual to change his or her behavior in order to adapt to cultural demands and expectations. Although this dialectical view contributed to Vygotsky’s emphasis on social mechanisms of development, an additional impetus came from the belief, shared by the majority of Soviet

94

Margarita Azmitia and Marion Brimutter

psychologists (e.g., Luria, Smirnov, Yendovitskaya, Zinchenko), that cognition is not an end in itself, but only a tool that allows individuals to become competent members of society. Inherent in this belief is the notion that to understand changes in cognition, one must understand the social context in which they take place. Although Vygotsky acknowledged the influence of social history and social institutions on development, he emphasized the relationship between interpersonal interaction and cognition. He proposed that children acquire cognitive skills while interactively solving problems with adults or more capable peers. Their more competent partners structure the problem-solving process in such a way that children are able to participate in increasingly complex parts of the solution, eventually becoming capable of solving the problem on their own. This transition from other- to self-regulation is the hallmark of Vygotsky’s approach. The zone of proximal development is another concept associated with Vygotsky. This zone is defined by the distance between the child’s independent problem-solving ability and his or her ability to solve problems with the help of others. Only interactions that occur within the confines of this zone are thought to lead to cognitive change. Even though two children may be quite similar in their independent problem-solving competence, they may differ in their ability to profit from interaction-that is, in the breadth of their zone of proximal development. Brown and her colleagues (e.g., Brown & Ferrara, 1985; Campione, Brown, Ferrara & Bryant, 1984) and Feurstein, Miller, Rand, and Jensen (1981) recently have undertaken a series of investigations designed to determine whether these differences in breadth of influence can be used as diagnostic tools in educational settings. In addition to exploring the relationship between individual differences in the breadth of the zone of proximal development and academic achievement, researchers (e.g., Brown & Ferrara, 1985; Bruner, 1985; Cole, 1985; Ellis & Rogoff, 1982; Forman & Cazden, 1985; Rogoff & Wertsch, 1984; Wood, 1980) have tried to operationalize the zone of proximal development and to assess empirically how interactions occurring within this zone promote selfregulation. These types of investigations have been very illuminating because they have focused on skill acquisition, and thus allow us to gain a better understanding of cognitive outcomes. Surprisingly, however, the majority of these researchers have not included assessments of children’s skills prior to interaction. If one does not establish the lower boundary of the zone of proximal development, that is, children’s independent competence prior to the interactive session, interactive effects are difficult to evaluate. Furthermore, available data are not adequate for evaluation of long-term consequences of mediated learning. As Dale (1986) commented during a colloquium, “We are ultimately not interested in producing the brightest first graders. We want

Social Influences on Cognition

95

to know how these (mediated) experiences influence children’s subsequent ability to learn.” Because of his focus on interpersonal agents of change, Vygotsky considered speech the most important mediator of cognitive development. Consequently, some of the researchers following in his tradition (e.g., Bruner, 1985; Goodsitt, Grady-Reitan, 8~ Perlmutter, in press; Ninio, 1983; Wertsch, McNarnee, McLane, & Budwig, 1980; Wood, 1980) have been documenting the nature of social agents’ speech to children and have validated its role in the development of self-regulation. However, although they have emphasized the role of the social agents’ speech in the growth of self-regulation, these researchers acknowledge the contribution of the child to changes in the agents’ regulation. The contribution of the child’s speech to internalization has been studied by observing changes in private speech (verbalizations that do not appear to be intended for others) as a function of development and of changes in task proficiency. This research strategy initially appeared promising because Vygotsky believed that private speech served a self-regulatory function and that its decline with age indexed the process through which the child transformed the social input into thought. Unfortunately, although several researchers (e.g., Kohlberg, Yaeger, & Hjertholm, 1968; Tinsley & Waters, 1982) attempted to identify developmental sequences in self-speech and to relate them to changes in task proficiency, these data have been inconsistent. Thus, the internalization process still remains a mystery.

2. Developmental Implications Like Mead, Vygotsky did not specify the nature of developmental changes in the nature of social interactions. However, unlike Mead, he did specify the process through which interaction benefits cognition, and as this article shows, the research generally has supported his views. Despite its greater refinement, Vygotsky’s model still has limitations. As mentioned, one serious limitation is the lack of an operational definition and empirical validation of the internalization process. In addition, some researchers (e.g., Kontos, 1983; Zaporozhets, 1969) have been concerned that he placed too much emphasis on social mechanisms (i.e., mediated activity) and failed to acknowledge the important contribution of solitary (i.e., unrnediated) activity. This overemphasis on social interaction can lead to the false assumption that interaction is always more conducive to learning than solitary behavior which, as this article shows, is not a tenable assumption. Finally, due to the emphasis on “other-regulation,” virtually no attempt has been made to specify the way in which the skills brought to the interaction by the child influence the interaction process. As we will argue, this omission is serious because it hampers

96

Matgarita Azmitia and Marion FPrlmutter

our ability to describe the nature of age-related changes in social interaction effects. C . STRUCTURAL PERSPECTIVE

1. Key Assumptions

In his earlier writings, working from a structural perspective, Piaget (1932) included social influences as one of four major factors in cognitive development. However, Piaget disagreed with Mead and Vygotsky’s contention that cognition originates in interaction. Rather, he proposed that to learn from interaction, children must already have the cognitive structures that allow them to assimilate social input. For Piaget, social interaction completes, but does not create, cognitive structures (Perret-Clermont & Brossard 1985). In support of this view, some researchers (e.g., J. P. Murray, 1974; Perret-Clermont, 1980) have shown that interaction benefits are more likely at transitional levels of performance than at initial stages of skill acquisition. Piaget (1932, 1968) proposed that social input becomes instrumental for cognition during the transition from preoperational to concrete operational thinking. As preoperational children become less egocentric, they become capable of recognizing discrepancies between their ideas and those of others and to use these intellectual conflicts to restructure their thinking, thus achieving a more advanced cognitive level. He also proposed that due to the relatively symmetrical nature of peer interaction (i.e., relatively little cognitive and social distance between peers), such interaction often would be more conducive to cognitive progress than would more asymmetric interactions (i.e., such as those occurring between children and adults or between children and peers of much higher cognitive or social status). This emphasis on symmetrical interactions derived, in part, from Piaget’s notion that children must be active in their own development; symmetric relationships are more likely to allow both members to participate in the problem-solving process. Symmetrical relations are also more likely to present new information in such a way that children can understand the nature of the discrepancy between their ideas and those of others and resolve the conflict productively. Much research (e.g., Kuhn, 1972; J. P. Murray, 1974; Perret-Clermont, 1980; lhriel, 1966) has provided empirical support for the superiority of symmetrical over asymmetrical interactions, but Doise and Mugny (1984) and Ellis and Rogoff (1986) have convincingly demonstrated that such a generalization may be overly simplistic. Their work demonstrated that in some settings, asymmetric interactions are more productive than symmetric interactions. Although in his later writings Piaget de-emphasized social mechanisms of cognitive growth, some of his followers (most notably Doise, Murray, Mugny,

Social Influences on Cognition

91

and Perret-Clermont) have emphasized the importance of these mechanisms within a Piagetian perspective. The bulk of the evidence suggests that at least by the elementary school years, conflict between individuals’ centrations can effectively promote cognitive growth. Moreover, the evidence suggests that social conflict is more conducive to learning than nonsocial conflict (e.g., the conflict between the child’s ideas and the properties of the task materials). The finding that progress achieved during interaction generalizes to subsequent individual performance and remains stable over time has done much to strengthen the case for the sociocognitive conflict mechanism proposed by Piaget.

2. Developmental Implications Piaget’s model was the first to incorporate developmental constraints on social facilitation. He argued that children would not benefit from interaction until the transition from the preoperational to the concrete operational stage, when they became capable of collaborating and considering alternatives. In support of this developmental claim, much of the research on the effects of peer interaction on preschool children’s cognition (e.g., Bearison, Magzamen, & Filardo, 1986; Perlmutter, Behrend, Kuo, & Muller, in press; PerretClermont, 1980) has failed to show that collaborative problem solving is more conducive to learning than solitary activity. However, before reaching the conclusion that social interaction does not contribute to learning in preschool children, one must entertain the possibility that less powerful effects, such as task engagement, may occur through early peer interaction or that other social agents, such as parents or older siblings, can enhance learning in this age group. As this article proposes, another possibility is that age is not the only relevant variable in the lack of facilitation obtained in studies of preschool children’s interactive problem solving. The complexity of the interactive task may play an equally important role. Although Piaget’s model has been one of the most influential, it still has not been addressed to some important questions. For example, it does not clearly indicate how a social mechanism can be incorporated as a cornerstone of a structural theory that emphasizes the role of the individual in cognitive development. Also, although social conflict has been shown to be more likely to provoke disequilibrium (and consequently to promote cognitive change) than nonsocial conflict, we still do not know why this is the case. We also have not identified the variables that make some social conflicts more productive than others. Perhaps, for example, children are capable of recognizing conflicting information long before they are inclined or able to act on this recognition, and the growth of the ability to utilize conflict influences the magnitude of social facilitation. Thus, for this framework, the problem

98

Margarita Azmitia and Marion Brimutter

involves incorporating a social mechanism into a structural theory of individual development and being more specific about the situations in which this mechanism fosters change. D. SOCIAL LEARNING PERSPECTIVE

1. Key Assumptions

Social learning approaches have emphasized the role of modeling in cognitive development. Children are assumed to acquire skills by observing and imitating others. Unlike the preceding theoretical perspectives, social learning theory does not require interaction or reciprocity between participants. It requires only that a child learn from observing an adult, a sibling, or a peer, although they need not be aware that their behavior is influencing the behavior of the observer. Because children prefer to imitate socially desirable models (Bandura, 1977), social reinforcement has been proposed as the impetus for imitation. One factor that makes some children more desirable models is their cognitive competence. In support of Bandura, Morrison and Kuhn (1983) found that as early as 3 years of age, children tend to imitate more competent peers (as opposed to children of similar or lower competence), and that this tendency increases with age. The fact that children learn through observation of others’ behavior does not necessarily indicate that they will imitate observed behaviors. Several factors, such as the consequences that befall the actor (reinforcement or punishment) and the characteristics of the actor (e-g., social status, perceived competence) influence children’s performance of learned behaviors. Also, children may perform behaviors acquired during observation in the absence of external reinforcement. Mowrer (1960) and Bandura (1977) have explained this finding by suggesting that after children have reached a certain level of task proficiency, imitation may be maintained through self-regulatory, or selfreinforcement mechanisms. For example, the satisfaction of solving a problem may motivate subsequent attempts at problem solving. Also, children can motivate themselves by repeating praise that others have used to motivate them in the past. Much controversy has arisen over whether the social learning approach adequately explains social influences on cognition. Research (e.g., Botvin & Murray, 1975; Perret-Clermont, 1980; Silverman & Geiringer, 1973) has shown, for example, that imitation cannot by itself explain the learning accrued from social interaction. In particular, imitation cannot explain why children often exhibit novel, more advanced behaviors following interaction and why experts d o not imitate the behaviors of the novices. Gewirtz (1969) and Miller and Dollard (1941) have argued that imitation is a supplemental learning process

Social Influences on Cognition

99

that cannot be used to explain novelty. Rather, imitation allows individuals to reorganize already available behavioral components into a different response or to apply already existing responses in a new context. Thus, on the one hand, critics of social learning theory must demonstrate that behaviors exhibited following interaction are truly novel (which may be difficult without longitudinal research), and on the other hand, social learning theorists may need to seek additional mechanisms of facilitation that will allow them to explain the acquisition of novel responses. In response to criticism that social learning theory cannot be used to explain why experts do not imitate novices, Zimmerman and Blom (1983) have argued that critics of social learning theory have interpreted imitation too literally. For example, Perret-Clermont (1980) argued that social learning theory predicts that conservers would be as likely to regress from exposure to a nonconserver as nonconservers to progress from exposure to a conserver. However, Bandura (1977) has long argued that imitation is selective, that children weigh the adequacy of information before imitating it, and, consequently, that competent models will be imitated more often than incompetent models. Although Zimmerman and Blom are correct in arguing that social learning theory can account for children’s preference for imitating more competent models, they must still explain how children recognize that a model is more competent. According to Piaget and Vygotsky, this ability to recognize one’s own incompetence and another’s greater competence, would only occur at transitional stages (Piaget) or within the zone of proximal development (Vygotsky) (H. W. Reese, personal communication, August 17, 1987). However, because social learning theory does not place developmental constraints on the ability to profit from interaction, researchers in that tradition must still clarify this issue. Zimmerman and Blom not only responded to criticisms of social learning theory, but also challenged the importance of the sociocognitive conflict mechanism proposed by Piaget. Specifically, they argued that interactive effects are explained better by the observational learning of rules than by the sociocognitive conflict mechanism proposed by Piaget. To test this hypothesis, they conducted a study in which nonconservers saw videotapes of two adults modeling conservation responses (rule-consistent condition), or modeling simple conflicting views (one conserver and one nonconserver), or inconsistent conflicting views (one models conservation, the other nonconservation, then they trade views). They argued that if social conflict induces cognitive conflict, the two conflicting conditions should foster more progress in conservation than the nonconflicting, rule-consistent condition. They also included measures of internal conflict (disequilibrium induced by cognitive conflict) to determine whether social conflict (conflict between individuals’ views) leads to internal conflict (disequilibrium). Their results showed that

100

Margarita Azmitia and Marion krltnutter

the children who were exposed to the rule-consistent condition progressed more than the children who were exposed to the conflicting conditions or than the children in the control condition. They argued that these data challenge Piaget’s views and provide support for the social learning position that conflict is neither necessary nor sufficient for cognitive development. Zimmerman and Blom’s (1983) study shows that imitation is a potentially viable mechanism for short-term social interaction effects. However, several methodological shortcomings in the design of their study and the lack of stability in learning accrued from interaction lead us to question whether their study can be used to challenge Piaget and whether imitation effects developmental (i.e., relatively permanent) cognitive change. For example, it is possible that the rule-consistent condition, inasmuch as the rule was different from the child’s rule, also instantiated a conflict situation and, thus, the sociocognitive conflict mechanism can be used to explain these findings. In addition, Cantor (1983) has questioned the adequacy with which internal conflict was measured and the breadth of the learning that occurred, given that the training and posttest items were very similar. Cantor also pointed out that the failure to find statistical differences between the rule-consistent and the conflict conditions in the delayed test suggests that the learning that occurred in the rule-consistent condition was not very stable. Finally, he suggested that because Zimmerman and Blom excluded transitional nonconservers, which is the group most likely to benefit from exposure to conflict (cf. Piaget, 1977), their test of Piaget’s mechanism was not adequate. However, F. B. Murray (1983) was correct in pointing out that whereas Zimmerman and Blom’s study may have limitations, it raises the question of whether conflict is necessary for cognitive growth. From his examination of Zimmerman and Blom’s and his own data (e.g., Botvin & Murray, 1975), Murray concluded that while conflict may be sufficient in some situations, it is not necessary for cognitive growth.

2. Developmental Implications Although imitation may play a role in learning, its role in effecting relatively permanent (i.e., developmental) changes in cognition has not been established. However, it is possible that imitation may emerge as an important mechanism in some situations, inasmuch as it allows individuals to reorganize their behavior and to decontextualize their skills. Like Mead’s and Vygotsky’s perspectives, social learning theory does not include specific developmental predictions. Because children begin to imitate others very early in life, this perspective requires the assumption that even very young infants benefit cognitively from interactions with others-possibly because this perspective does not require reciprocity in the earliest mechanism for social interaction

Social Influences on Cognition

101

effects. Although this assumption may be accurate, it does not allow us to address the question of the limits of social influences on cognition and how these limits change with age. E. CONTEMPORARY PERSPECTIVES

Although the importance of social mechanisms of cognitive growth was recognized at least by the early 20th century, these ideas did not importantly influence research until much later. The unavailability of translations of Piaget’s and Vygotsky’s work before the late 1960s surely contributed to this delay, but even Piaget’s followers in the Genevan school did not actively pursue sociocognitive issues until the 1970s. Rather, cognitive developmentalists who followed Piaget were primarily influenced by his late formulations, which did not attribute a special role to social interaction. Moreover, much prior work on cognitive development was dominated by the information-processing perspective, in which, to this day, social influences on cognition are largely ignored. A number of different factors appear to have contributed to the emergence of interest in social mechanisms of cognition. First, mushrooming crosscultural research has demonstrated the importance of context for cognition (see Laboratory of Comparative Human Cognition, 1983; Rogoff, 1982; D. A. Wagner & Stevenson, 1982 for reviews). Second, an accumulation of documentation indicates considerable variability in cognitive development (e.g., Brown, Bransford, Ferrara & Campione, 1983; Fischer, 1980; Super, 1980) that is not easily handled by leading models of cognitive development. Third, many of the skills used by children to solve tasks in their everyday social settings are now realized to be inadequately captured by laboratory tasks (e.g., Perlmutter, 1980; Rogoff & Lave, 1984). For example, although in the everyday environment children can ask others for assistance, they are not permitted to do so in the laboratory. In fact, very few children fail to seek help when they are confronted with a problem that exceeds their capabilities. Fourth, the political climate of the 1970s was conducive to a search for social factors that could enhance achievement (e.g., the Head Start movement). F. SUMMARY

In this section, we briefly summarized leading theoretical perspectives on social influences on cognition and provided a historical overview of the factors that have led to the current interest in this problem. The discussion indicates that sociocognitive conflict (Piaget) and internalization of regulation (Vygotsky) are the two dominant mechanisms proposed to account for social facilitation of cognition. Piaget’s conceptualization of sociocognitive conflict may

102

Matgarita Azmitia and Marion Rrlmutter

be viewed as a provocation to equilibration. After a state of disequilibrium has occurred, the child is assumed to resolve the conflict on his or her own, although the resolution may be achieved through the child’s negotiation of a compromise position (cf. Youniss, 1980). In contrast, Vygotsky’s conceptualization of social influence on cognition incorporates the substance of the interaction as crucial to cognitive change. The regulation provided by another eventually becomes internalized and regulates the child’s cognition. Note that the child does not internalize an exact replica of the social agents’ guidance. Rather, he or she actively modifies the input, transforming it into a tool that can be more powerful than the original tool. On the surface, the only difference between Vygotsky’s and Piaget’s perspectives may seem to be that the former incorporates the substance of the interaction into the learning process, and the latter does not. The similarity between both perspectives is further highlighted by noting that both of their mechanisms imply some level of conflict. However, the two perspectives differ in at least three important ways. First, Vygotsky allocated a more important role to social interaction (all skills originate in interaction) than Piaget (interaction is one source of skills, and then only at transitional points of skill acquisition). Thus, for Vygotsky, the problem is to specify how the social is translated into the individual. As mentioned, for Piaget, the opposite is the case; that is, researchers in this tradition must incorporate a social mechanism into a theory of individual development and explain why social conflict is more conducive to change than nonsocial conflict. Second, Piaget, but not Vygotsky, places developmental constraints on social influences-children will not benefit from interaction until the concrete operational period. Third, although Piaget considered that symmetric interactions are more conducive to learning, Vygotsky emphasized the contribution of asymmetric interactions. Imitation, although prematurely disregarded, is likely to reemerge as the mechanism for some social effects on cognition. Because observational learning requires minimal, if any, reciprocity, this mechanism may be particularly powerful during the early years, when children lack the social and cognitive skills needed to sustain collaboration. However, this perspective still needs to explain novel behaviors that emerge following the interaction. Traditionally, social learning theorists have explained novelty by arguing that because imitation is often inaccurate, the resultant behavior can differ from the original behavior. However, although this argument would explain distortions, it does not explain novelty. Even if one accepts Bandura’s suggestion that novelty arises from the individual’s combination of observed behaviors to produce a response, the emergence of behaviors that bear no resemblance to interactive behaviors, or combinations thereof, remain unexplained. Social learning theorists must also explain how children come to realize that another’s strategy is more efficient -that is, why novices imitate experts but experts do not imitate novices.

Social Influences on Cognition

103

In sum, although historical and contemporary approaches have implicitly or explicitly dealt with developmental issues, we presently lack a coherent picture of how development moderates social influences on cognition. Much controversy has centered around which mechanism best explains social facilitation. Although the resolution of this controversy may be desirable from a theoretical standpoint, this controversy has directed attention away from developmental questions concerning these mechanisms. These mechanisms may be complementary, and each of them may be salient at different points of development. For example, for reasons discussed previously, very young children may learn more through imitation than through discussion of conflicting ideas because they lack the skills to resolve conflict productively. Thus, any attempt to document the contribution of conflict, disagreements, and discussion to cognitive development during the preschool years may be futile, although one can certainly attempt to discover precursors of the ability to resolve conflicts and orchestrate discussion. The following section reviews empirical evidence of social influence on cognition, with special emphasis on the mechanisms that have been proposed to explain these effects.

111. Evidence for the Impact of Social Agents A.

RELATIONSHIPS AND COGNITIVE DEVELOPMENT

I. Function of Relationships in Cognitive Development

Hartup (1985) has suggested that relationships serve three main functions in cognitive development. First, they are contexts in which skills emerge. Second, they provide information about appropriate behaviors and emotional support needed to explore new horizons. Third, relationships are the forerunners of other relationships. We would broaden this claim, and suggest that social interaction in general contributes to cognitive development in these ways. Nevertheless, we acknowledge that although any individual can serve as a social agent and influence another’s learning, such influence may be more likely in the context of relationships, that is, with a well known other such as a parent, sibling, or friend. Relationships may be particularly good contexts for learning and development because well known others are aware of each one’s strengths and weaknesses and thus are able to tailor guidance accordingly. Also, individuals feel more comfortable asking questions and offering guidance if they know each other. Finally, because of their previous interaction history, well known individuals do not need to devote much energy to the social demands of interaction and can thus concentrate their resources on the cognitive demands of the problem. As will be seen, few studies have evaluated the contribution of

I04

Margarif0Azmitia and Marion Arlmurter

relationships to cognitive development. However, the available ones are supportive of this contribution. For example, the quality of children’s attachment has been related to exploration and problem solving (Sroufe & Flesson, 1986) and friends derive more benefits from collaborative problem solving than acquaintances (J. Nelson & Aboud, 1985; Newcomb & Brady, 1982).

2. The Unique Contributions of Parents, Siblings, and Peers Vygotsky emphasized the contribution of asymmetric (e.g., parent-child, older sibling-child, expert child-novice child) relationships to cognitive development. In contrast, Piaget emphasized the contribution of symmetric (e.g., friends, children of similar social and cognitive status) relationships. An important point, however, is that symmetry is a matter of degree, and that both Piaget and his followers recognized that input from a child that is slightly more advanced can foster learning. Contrary to current assertions, neither theoretician considered peer influence to be more important than parental influence. Rather, they argued that these sources of influence represent two distinct but equally important avenues of development. We would argue that developmental changes in the child, as well as age structuring in the society, contribute to the relative salience of these different kinds of social influence. Children appear to be born with biological predispositions that facilitate social interaction. For example, they show preference for the features of the human face (e.g., contour, contrast, movement) and speech. However, early social interactions are managed by the social agent. For example, Cazden (1979, cited by Rogoff & Gardner, 1984) found that although mothers initially will accept any response from their infant, often even answering their own questions to keep the interaction going, after the infant is about 7 months old, they become more demanding, accepting only the responses they consider appropriate (e.g., vocalic babbles, as opposed to any sound). Although changes in the mother’s demands contribute to increases in the infant’s active participation in interactions, the child’s desire to emulate and please the caregiver (cf. Wood, 1980), and changes in his or her cognitive and social skills, also contribute to this increase. Toward the end of the first year of life, the child begins to move toward decentration-that is, to develop a concept of self and to recognize that objects in the environment are independent agents (Piaget, 1954). This cognitive achievement influences the child’s social development because it allows him or her to recognize the reciprocity and intentionality that is inherent in social interaction. In turn, this achievement allows the child to take a more active role in interaction and to respect the needs and expectations of others (Brownell & Brown, 1985). These achievements mark the onset of socialization, for which the responsibility initially falls on the parents. Parents’ goal in socialization is to commmunicate the skills needed to become functional, productive members of society. By

Social Influences on Cognition

105

definition, the parent-child relationship is asymmetric because the parent is the expert on society’s rules and expectations. Although parents often seek their children’s input, ultimately any conflict between their views will be resolved in favor of the parent because the parent can appeal to an external agent-society. Thus, in parent-child relationships, parents are largely in control of the interaction (Youniss, 1980). Given the nature of the parent-child relationship, children often conform to rules that they do not fully understand. A major challenge to this conformity occurs when children begin interacting with peers. When the child is confronted with a peer who disagrees with his or her views, conflict ensues. Although children can appeal to an external arbiter (society) to resolve their differences (e.g., boys do not wear dresses), often this appeal is not sufficient to resolve the conflict. In such a situation, children are forced to negotiate a compromise-that is, to co-construct a new understanding (cf. Sullivan, 1953; Youniss, 1980). Part of this understanding involves realizing that because rules are created through group discussion, they can be changed in certain situations (Piaget, 1932). Because this co-construction is a cooperative process, every participant becomes an agent of his or her own and others’ development. Only through this co-construction of ideas do children come to understand fully the links between their behavior and that of others (Youniss, 1980). Although peer interactions may be characterized by greater equality or symmetry than parent-child relations, equality is not a given in peer relations. Many interactions between peers, especially those occurring between acquaintances (as opposed to friends) are marked by children’s failure to treat each other as equals-that is, by coercion, humiliation, and general lack of sensitivity to each others’ needs (Krappman & Oswald, 1987). Consequently, not all conflictual interactions between peers result in the co-construction of a new understanding. Despite this caveat, the cooperative nature of peer relations (as opposed to the control inherent in parent-child relations) may allow children to be more active in the learning process. For example, in their study of the effectiveness of child versus adults tutors, Ellis and Rogoff (1986) found that the learner tended to be more compliant and shy when working with an adult than when working with another child. When working with another child, the learner actively sought more participation in the problem-solving process and was more likely to seek guidance and clarification. This latter tendency may reflect the fact that adult tutors spontaneously involved the child in the problem solving process and provided guidance and clarification.

3. The Development of Interaction Skills As discussed, infants begin to take a more active role in interaction during the second half of the first year of life. Not only do they begin to initiate

106

Margarita Azmitia and Marion Arlmutter

and to maintain interactions by using eye contact, facial expressions, and objects to communicate their intent, but they also begin to use adults as tools to attain their goals (e.g., obtaining a toy that is out of reach, transporting them to a desired locale). The success of these interactions depends on infants’ and adults’ ability to read each others’ signals; that is, adults often must generate hypotheses about the referent of nonlinguistic (and even some linguistic) signals, and infants must find a way to indicate which one of the adults’ hypotheses are correct. By the end of the first year of life, it appears that infants and adults have negotiated a sophisticated “conversation of gestures” (Rogoff, Mistry, Radziszewska, & Germond, in press). Historical changes (e.g., exposure to day care) have increased very young children’s exposure to peers. Thus, many researchers (e.g., Brenner & Mueller, 1982; Brownell & Brown, 1985; Mueller & Brenner, 1977; Renninger & Morgan, 1986) have explored the possibility that peers begin to influence children’s cognitive development at an earlier age than had been previously assumed. Hartup (1985) has suggested that prior to the emergence of combinatorial skills at the end of the second year, interactions between same skilled peers are rare because combinatorial skill is necessary for reciprocity. After combinatorial skills emerge, same-skilled peers become capable of maintaining interactions, but these interactions undergo quantitative and qualitative changes during the preschool years. For example, interactions between young children are brief and infrequent (e.g., Brenner & Mueller, 1982; Renninger & Morgan, 1986). They also center around object exchanges and only later become augmented by gestures and verbalizations across distances (Brownell & Brown, 1985). Consequently, due to cognitive and social limitations, interactions between young peers may not be very productive. For example, children must first be capable of reciprocity and then be able to attend to and to evaluate others’ views and to communicate their own effectively. They must also be capable of negotiating a satisfactory division of labor and of seeking and giving help when needed. Finally, they must learn to resolve conflicts productively. Youniss observed that because preschool children are not very proficient in these skills, their conflicts often stymie or dissolve the interaction. In contrast, interactions between parents and young children can be productive because parents have the skill needed to compensate for the child’s limitations and thus are able to support his or her behavior during interaction. The emergence of combinatorial skills, however, is not the only variable mediating increases in the sophistication of peer interactions. Parent-child interactions also are more successful than peer interactions because their long interaction history has allowed them to learn how to read each others’ cues. Killen (1987) has suggested that the more sophisticated interactions reported by researchers studying parent-child interactions relative to those reported

Social Influences on Cognition

107

by researchers studying peer interactions are not only due to the parents’ greater skill in supporting their children’s behavior, but also to their longer interaction history. Because the majority of researchers studying peer interactions have focused on interactions between unfamiliar, or at least only slightly acquainted peers, we may have underestimated the sophistication of early peer interactions. The research that is discussed here lends support to Killen’s hypotheses: Even at an early age, interactions between friends show much of the sophistication evident in parent-child interactions. In addition to changes in the length, quality, and frequency of peer interactions during the preschool years, children’s criteria for selecting partners also changes. Initially, activities dictate children’s choice of partners, but at the end of the preschool years children begin to base their choices on their partner’s dispositional qualities (Furman, 1982). Also, although preschool children tend to base their bids for help on the helper’s perceived competence (thus preferring to seek out adults), elementary school children tend to base their bids for help on their relationship to the helper (thus preferring to seek out friends, regardless of their competence) (Nelson-Le Gall & Gummerman, 1984). Apparently, then, the shift in the salience of parents and peers across development is mediated not only by increasing exposure to peers, but also by changes in the child’s cognitive and social skills and his or her selection of partners. Investigations of social influences on cognition have been influenced by the foregoing developmental considerations. For example, studies of parentchild interaction constitute the bulk of research on the contribution of social influences to cognitive development during the first 5 years of life. In contrast, reseachers studying social influences on cognition during the elementary school years have focused largely on peer interactions. In the next sections, we review some of the research on the contribution of parents, peers, and - more recently- siblings, to children’s cognitive development. We also highlight the complementary nature of these relationships. B.

PARENT-CHILD INTERACTION

Numerous studies (e.g., Behrend, Rosengren, & Perlmutter, 1986; Debache, 1983; DeLoache & Plaetzer, 1985; Eder, Parks, Todd, Robb, & Perlmutter, 1986; Goodsitt et af.,in press; Kontos, 1983; Ninio, 1983; Ninio & Snow, 1985; Radin, 1982; Rogoff & Gardner, 1984; Rosengren, Behrend, & Perlmutter, 1985; Wertsch, 1979; Wertsch etul., 1984; Wood, 1980) have provided suggestive evidence that parent-child interaction may influence cognition and that some parents fine-tune interactions in ways that facilitate their children’s acquisition of skills. This fine-tuning, or scuflofding process, has been found in a wide range of contexts, from those involving academic skills (e.g., Ninio,

108

Margarita Azmitia and Marion Arlmutter

1983; Saxe et al., 1984; Wood, 1980) to those involving more informal skills, such as play and weaving (DeLoache & Plaetzer, 1985; Greenfield & Lave, 1982). However, although parents frequently scaffold their children’s performance, after the first year of life, children begin to take a more active role in the interaction. Moreover, during the late preschool years, children may actively resist their parents’ attempts to structure the interaction if they are working on a familiar task or if a peer is available. Although the studies just mentioned suggest that parent-child interaction may influence cognitive development, the literature provides very little evidence for a causal connection between parent-child interaction and cognitive growth. This lack of evidence does not eliminate the possibility of a causal connection. The problem is that the design of most studies obviates causal statements. For example, most of these studies have been correlational, and thus allow only the conclusion that social influence and cognition are associated. Also, few studies have included an assessment of children’s abilities prior to and following the interaction. Thus, we cannot test the hypothesis that interaction fosters cognitive growth. Finally, the preponderance of cross-sectional designs limits our ability to make strong statements about the relationship between age-related changes in parent-child interactions and age-related changes in cognition.

1. The Contribution of Scaffolding to Cognition Wood and his colleagues (Wood, Bruner, & Ross, 1978; Wood & Middleton, 1975; Wood, Wood, & Middleton, 1978) have conducted a program of research that is one of the few that has included outcome measures and empirically manipulated interactive variables assumed to mediate changes in children’s competence. They performed fine-grained analyses of how parents structure interaction to promote their children’s progress and found that some patterns of interaction are more helpful than others. For example, parents who structure their assistance according to their child’s need for guidance-that is, intervene only when the child makes a n error or is uncertain of how to proceed-are more likely to increase their child’s competence during and following the interaction than are parents who do not show such sensitivity. Consistent with this conclusion, other researchers have shown that parents of young children tend to take a more active role in monitoring the interaction than do parents of older children (Rogoff & Gardner, 1984; Wertsch et al., 1980), using a greater number of commands (Laboratory of Comparative Human Cognition, 1983), and emphasizing skills that are consonant with their child’s competence. For example, when reading books to toddlers, parents emphasize labels and basic book-reading skills like page-turning, but when reading to older children they emphasize more sophisticated skills like information integration and memory (Goodsitt et al., in press; Ninio & Snow, 1985).

Social Influences on Cognition

109

An important question that has not been addressed well is how parents know to alter their guidance. Although the research of Wood and his colleagues suggests that children’s task performance provides one source of information, other sources may also be involved. For example, in addition to verbal requests for assistance, nonverbal signals can be important cues (Rogoff 8z Gardner, 1984). Children’s facial expressions may serve as cues to promote or inhibit parental involvement, for instance, a surprised expression may promote involvement and a look of intense concentration may inhibit it. Also, children’s glances may help regulate interactions by signaling their need for aid (e.g., Winegar, 1986) or their interest in initiating or terminating the exchange (e.g., Rogoff, Malkin, gL Gilbride, 1984; Winegar, 1986). By identifying the cues that mediate the moment-to-moment dynamics of interaction, we might be able to define the mechanisms of facilitation better. Although parental scaffolding appears to facilitate cognitive development, it is still unclear whether scaffolding is the most beneficial interaction style. Some researchers (e.g., Reeve, 1987; Wood, 1980) have shown that scaffolding promotes greater learning than other teaching strategies, but others (e.g., Goncu and Rogoff, 1987) have found that observing an adult solve a problem (i.e., modeling) and providing only occasional guidance can be as effective as scaffolding. One problem standing in the way of resolving this controversy is that scaffolding and other teaching strategies, such as demonstration, share many behaviors. For example, scaffolding incorporates demonstration. Consequently, we still d o not know which unique features of these interaction styles are responsible for learning. Also, as we argue here, the usefulness of these techniques may be determined, at least in part, by the task context. Demonstration may be sufficient for teaching construction skills, but not for teaching more sophisticated skills such as planning and math.

2. Parents’ Contribution to Children’s Generalization of Skills One of the most important questions faced by researchers in cognitive development is how individuals generalize their skills to new contexts. Rozin (1976), for example, argued that much of cognitive development is characterized by skills becoming increasingly accessible to consciousness, and, in turn, to new situations. The publication of seminal works such as Donaldson’s (1978), Flavell, Botkin, Fry, Wright, and Jarvis’s (1968), and Gelman’s (1978), which demonstrated that preoperational children exhibit conservation of number and nonegocentric perspective taking in some situations, raised the question of why children can apply their skills to one problem and yet fail to apply them to another problem that requires the same skills. Ultimately, the developmental issue raised by these data is how individuals learn to categorize some events and problems as similar and others as different. It is possible that through social interaction, parents may foster their child’s

110

Margarita Azmitia and Marion krlmutter

cognitive competency by formulating explicit connections between present problem-solving contexts and the child’s previous experience. The data (e.g., Ellis & Rogoff, 1982; Laboratory of Comparative Human Cognition, 1983) suggest that parents often draw their children’s attention to related and redundant events. By transforming the new into the familiar, parents help their children decontextualize their strategies, which is an important feature of cognitive competence (Brown et al., 1983; Laboratory of Comparative Human Cognition, 1983; Rogoff & Gardner, 1984). If such connections between the old and the new are not made explicit by others, even adults have difficulty transferring skills between similar problems (Gick & Holyoak, 1980, 1983). Social input may also accelerate children’s adoption of accidentally discovered strategies. Kuhn and Phelps (1982) have shown, for example, that although individuals frequently discover more advanced problem-solving strategies, several encounters with them are needed before they realize the significance of the strategies. By pointing out their significance, parents can help children incorporate these strategies into their repertoires. One important question is whether all of these functions are also served by peer interaction. To the best of our knowledge, there is no data on this question.

3. Some Final Reservations about Cognition and Parent-Child Interaction The previous discussion highlighted the potential contribution of parentchild interaction to early cognitive development. However, the evidence has not always supported the view that parent-child interaction facilitates early cognitive development. For example, attempts to map parental strategies to children’s behaviors have not always been successful (Davis & Lange, 1973; Kontos, 1983; Ratner, 1980), and gains seen during interaction do not always generalize to children’s subsequent individual performance. Another problem is that giving children opportunities to practice solving problems on their own can be as effective in enhancing problem-solving as parent-child interaction (Kontos, 1983; Sylva, Bruner, & Genova, 1976). These challenges are not so serious, however, because they mitigate only the strong claim that all skills originate in social interaction, not the more viable claim that social interaction can be useful in promoting some cognitive skills. Very likely, if a skill is crucial for children’s instrumental competence, they will have multiple opportunities to acquire it, and these opportunities will involve both individual and social contexts. The failure to find direct connections between interactive behaviors and children’s subsequent activities is also to be expected if time is required for interactive benefits to materialize. For example, Kuhn and Phelps (1982) found that strategy acquisition is characterized by highly variable application of the

Social Influences on Cognition

111

strategy, and thus, using only one posttest to assess acquisition can lead researchers to underestimate (or alternatively, overestimate) the degree of transfer of strategies from the interactive to the individual situation. Also, Ratner (1984) found that although parents’ memory demands were unrelated to their 2-year-old’s memorial competence, they were positively and significantly correlated to their memorial proficiency at age 3. Finally, expecting to find a direct parallel between interactive and solitary behaviors may be unreasonable. Both Vygotsky and Piaget argued that internalization is a constructive process, and thus the skill that emerges may not bear close resemblance to the original interactive skill. Only social learning approaches, which explain social effects through imitation, would predict that a skill that bears close resemblance to the interactive skill would emerge in children’s subsequent behavior. Although this explanation is plausible, in order to avoid circularity, it must be validated empirically. A final reason for the disconfimation of parental benefits in some interactive contexts may be that only certain skills benefit from interaction. Executive or metacognitive skills have been proposed as the most likely candidates to be transmitted through social interaction (Hartup, 1985). Such skills may be particularly important because they are necessary for self-regulation (Vygotsky, 1962). They also seem to embody the essence of the fine-tuned interactions that foster the transition from other- to self-regulation (Hartup, 1985). For example, to be a successful problem solver, the child needs to be able to define the problem, select a strategy, and evaluate the adequacy of the solution. In support of this proposal, Moss (1984) found that mothers emphasized executive skills such as goal structuring, reality testing, and predicting consequences while assisting their children in problem solving, and Rosengren et al. (1985) found that parents emphasized general problem-solving strategies, as opposed to specific task behaviors, during interactions with older preschool children. The fact that general metacognitive skills are often unrelated or only indirectly related to speclfic task behaviors (Cavanaugh & Perlmutter, 1982; Wellman, 1983) may account for the low correlations between children’s interactive behaviors and their subsequent individual performance.

4. Summary Although some of the initial claims made about the influence of parentchild interaction on early cognitive development need to be tempered to accommodate empirical challenges, the evidence generally suggests that parents may play an important role in early cognitive development. However, we suggest that attention must also be paid to the strategies children use to regulate their parents’ guidance. Future researchers should attempt to (1) establish causal connections between parent-child interaction and cognitive growth and

112

Morgorito Azmitia and Marion krlmutter

(2) discern whether some skills are more likely than others to benefit from interactive contexts. In addition, although these data are still fairly sparse on this issue, we think that attention to nonverbal regulators of interaction may be beneficial. C. PEER INTERACTION

Although some evidence suggests that peer interaction influences cognitive development, the same problems that prevented us from making causal statements about the relationship between parent-child interaction and cognitive development apply here. That is, most studies have been correlational and have not included pretests and outcome measures. 1. Collaborative Problem Solving during the Preschool Years For reasons discussed previously, very few researchers have explored the contribution of peer interaction to preschool children’s cognitive development. Some studies have shown that preschool children can solve problems interactively (e.g., Brownell, 1982; Cooper, 1980) and that they are more engaged and enjoy joint problem solving more than solitary problem solving (eg., Perlmutter et al., in press). In addition, some types of peer interactions (e.g., interactions between experts and novices) are more conducive to learning than solitary activity (Azmitia, 1988), and interactive gains can generalize to preschool children’s subsequent individual performance (Azmitia, 1988). Nevertheless, many studies have failed to produce evidence that for preschool children interactive contexts promote more productive activity (Perlmutter et al., in press) or greater learning than solitary situations (Bearison et al., 1986; Gauvain & Rogoff, 1985; Perret-Clermont, 1980). These findings may support Piaget’s claim that there are developmental constraints on peer interaction benefits. However, alternative explanations are possible. For example, the studies that have yielded support for interactive benefits have involved tasks that are quite familiar to preschool children (e.g., building with blocks, simple combinations of materials), and thus, children may have entered the problem-solving situation with some of the prerequisite cognitive skills needed to solve the problem. Work with older children (e.g., Doise & Mugny, 1984; Ellis & Rogoff, 1982; J. P. Murray, 1974; Syc, 1986) has shown that interactive benefits are unlikely when children are complete novices at a task, because the cognitive demands of finding the solution and the social demands of coordinating interaction prove too ovenvhelming. With difficult tasks, even if progress occurs during interaction, the cognitive gains may be short-lived, because after a few weeks, children may return to their starting level of competence (Field, 1981). Thus, when the task

Social Influences on Cognition

113

is fairly difficult, only parents, expert peers, or older siblings may be able to support the child’s problem-solving behavior and to foster cognitive growth. The importance of considering task familiarity when assessing interactive benefits is highlighted further by the observation that the studies in which social interaction has not facilitated performance have involved either relatively difficult tasks, such as solving conservation problems (Bearison et al., 1986; Perret-Clermont, 1980), planning an efficient shopping route (Gauvain & Rogoff, 1985), and learning a programming language (Perlmutter et al., in press), or tasks that were initially unfamiliar to both children (Perlmutter et a/., in press). The importance of this factor is also supported by Flavell et a/.’s (1968) elegant demonstration of the roles of task familiarity and complexity in children’s ability to take the perspective of others into account and by Donaldson’s (1978), Gelman’s (1978), and Pratt, Scribner, and Cole’s (1977) finding that when tasks are constructed to fit the competence of preschool children, they show remarkable social and cognitive sophistication. Research also provides some suggestion (e.g., Azmitia, 1988) that if at least one of the members of a dyad is familiar with the task, he or she will be able to manage the interaction such that the novice becomes increasingly competent. Thus, an important goal for future research is systematic study of the relationship between task difficulty and the magnitude and types of effects accrued from preschool children’s collaborative problem solving. Such research is needed to determine whether age, task difficulty, or both are the relevant variables. More importantly, although the benefits of interaction may not be evident by criteria that are focused only on task accuracy, other measures may reveal interactive benefits. For example, although they failed to find differences beteen dyads’ and singletons’ task performance, Perlmutter et al. (in press) found that dyads were more engaged in the task and that they enjoyed the task more. Also, Gauvain and Rogoff (1985) found that dyads considered a wider variety of alternatives in planning a shopping trip, and Azmitia (1988) found that low-ability dyads spent more time on task than did low-ability singletons. It can be argued that greater engagement, enjoyment, consideration of alternatives, and persistence may set the stage for subsequent cognitive benefits, although this suggestion requires empirical confirmation. Longitudinal data would be particularly valuable for validation of this hypothesis.

2. Peer Interaction during the Elementary School Years Although the exploration of preschool children’s problem-solving interactions has been limited, support for the benefits of peer interaction has been obtained for the elementary school years (e.g., Allen, 1973; Cazden, Cox,

114

Margarita Azmitia and Marion krlmucter

Dickinson, Steinberg, & Stone, 1979; Cooper, Ayers-Lopez, & Marquis, 1982; Doise & Mugny, 1984; Forman & Cazden, 1985; Hawkins, Homolski, & Heide, 1984; Perlmutter et a/., in press; Perret-Clermont, 1980; Skon, Johnson, & Johnson, 1981). However, because many of these studies have involved peer tutoring and not the more informal problem-solving interactions that have characterized studies involving preschool children, we might have overestimated the contribution of peer interaction in general to cognitive growth. Unlike preschool children, elementary school children have acquired the cognitive and social skills needed to support collaboration and therefore are capable of productive interactions in a wider range of situations. Collaboration may be impeded only in cases where the task is too difficult (e.g., Syc, 1986) or when the advantages of collaborating on a task are not obvious (e.g., Renshaw & Carton, in press). Some educators (Feurstein et al., 1981; Johnson, Johnson, & Roy, 1984) also have argued that because Western school systems encourage individual work, children may need to be trained on collaborative skills (e.g., division of labor, conflict resolution) for collaboration to be successful. If such training is not provided, children may be quite hesitant to work together and may actually unintentionally sabotage the group goal and end up working independently. Until recently, Piaget’s framework had dominated research on the effects of peer interaction, and thus the tasks he used (e.g., conservation, chemical solutions) have been widely used to investigate interactive benefits. In addition to their theoretical significance, these tasks have been widely used because they allow developmental evaluation of children’s performance and changes thereof. However, because they are relevant only for a very restricted age range, studies involving these tasks have not provided much information about developmental changes in interactive problem solving during the elementary school years. Perhaps the most extensive investigations within the Piagetian tradition have been carried out by Doise, Mugny, and Perret-Clermont (Doise & Mugny, 1979, 1984; Doise, Mugny, & Perrett-Clermont, 1975, 1976; Mugny & Doise, 1978; Perrett-Clermont, 1980; Perret-Clermont & Brossard, 1985), European researchers from the Genevan school, and Bearison (1982; Bearison & Cassel, 1975; Bearison et al., 1986), Murray (Ames & Murray, 1982; Botvin & Murray, 1975; F. B. Murray, 1981, 1982), and Tbriel(l966, 1983) in the United States. Their research has yielded great advances, first by showing the superiority of interactive over solitary problem solving, and subsequently by identifying some of the features of interaction that promote or hinder cognitive proficiency. The importance of this accomplishment should not be minimized because it has allowed us to go from generally knowing that interaction can instigate cognitive development to beginning to understand why this is the case. As this chapter indicates, however, many gaps in our knowledge remain.

Social Influences on Cognition

115

At a simple level, providing a potentially interactive context does not guarantee that children will work together (Doise & Mugny, 1984; Hawkins et al., 1984; Renshaw & Garton, in press). Unless children include collaboration in their definition of the task goals, solitary work will be as beneficial as interaction because, in effect, children are working independently in both situations. As mentioned, if the task is too difficult, very young children may not be able to maintain collaboration, and if the task does not afford an obvious advantage to a collaborative solution, even older children may prefer to work independently. In addition, Bearison et al. (1986) found that too much conflict can actually inhibit learning; Zimmerman and Blom (1983) found that social conflict does not necessarily entail internal (cognitive) conflict; and Damon (1983) demonstrated that unless children can successfully reach an agreement about the solution, progress is unlikely (but see Glachan & Light, 1982; Perret-Clermont, 1980; Skon et al., 1981). The fact that the ability to reach agreement and make decisions is limited in young children may contribute to the lack of facilitation accrued from some preschool children’s peer interactions. Rvo other findings that have emerged from this research are more controversial. The first concerns the level of expertise needed to profit from interaction. The second concerns the relationship between interaction styles and interactive benefits. Because expertise and interaction atyle are likely to interact with age, we first discuss the general effects attributed to these factors and then discuss how these effects may change across development.

3. The Relation between Expertise and Learning Concerning the level of expertise needed to profit from interaction, educators (e.g., Allen, 1973; Johnson et al., 1984; Skon et al., 1981) have suggested that progress occurs independently of expertise-that is, that highability children benefit as much as low-ability children from social interaction. However, other researchers (e.g., Doise & Mugny, 1984; J. P. Murray, 1974; Perret-Clermont, 1980) have challenged this claim and have shown that lowability children must have some of the prerequisite skills for the problem’s solution (i.e., be transitional conservers as opposed to nonconservers) in order to benefit from interaction. Given these prerequisites, novices benefit more than experts (Doise & Mugny, 1984). Doise and Mugny explained this finding by alluding to Piaget’s argument that social interaction is most beneficial during the initial genesis of skills, in which it leads to disequilibrium. After such disequilibrium has occurred, the child retires to the privacy of his or her mind, so to speak, to accommodate the discrepant information. One possible explanation for the inconsistency in the results of studies in which expertise has been manipulated concerns methodological differences. Educators provide extensive training in interactive skills prior to the experimental

116

Margarita Azmitia and Marion Rrtmutter

sessions, but such has not been the case in research that has supported the necessity of intellectual prerequisites. These studies (e.g., Doise & Mugny, 1984; Syc, 1986) have shown that if the task is too difficult, the cognitive demands of the task may prove so overwhelming to low-ability children that if they are not schooled in how to resolve conflicts or to negotiate and maintain collaboration, the interaction will disintegrate, children will opt for parallel working styles, and no progress will occur. A second reason for the discrepancy concerns differences in the operational definition of expertise. Expertise defined as general academic achievement generally is independent of interactive effects (e.g., Skon ef ul., 1981), but expertise defined as skills specifically related to the task appears to predict interactive benefits (e.g., Doise & Mugny, 1984, J. P. Murray, 1974). This finding is reminiscent of the large body of work (for reviews, see Cavanaugh & Perlmutter, 1982; Wellman, 1983) that showed that although general metamemorial knowledge is not a good predictor of memory performance, when metamemorial knowledge that is specific to the memory task is correlated with memory performance, a strong positive correlation emerges between these two factors. Differences in operational definitions of expertise are not sufficient to account for the discrepancies, however, because some researchers who have defined expertise in terms of task proficiency (e.g., Ames & Murray, 1982; Glachan & Light, 1982) have found that low-ability dyads can improve more than low-ability singletons. A special feature of this research has been that novices with opposite centrations (e.g., in a length conservation task, one child believes that a particular line is longer while the other child believes that it is shorter) have been paired. In addition, children have been told that they must agree on the solution. This type of situation maximizes the chance that children will experience a conflict of centrations and thus optimally sets the stage for disequilibrium. It thus appears that both generalizations-that expertise does not influence interactive benefits and its converse that expertise does influence interactive benefits-appear overly simplistic. In situations where children’s deficiencies complement each other or children have received extensive training in how to collaborate, expertise may not constrain progress. Note, however, that both of these situations are fairly artificial. Thus, the findings obtained from training studies or studies pairing children with complementary skills must be interpreted with caution. These findings may tell us much about the potential of interactive contexts for learning, but may tell us little about the natural developmental process within which interaction effects operate. Although the preceding discussion should have highlighted the importance of considering expertise when predicting the success of interactive problem solving, it does not allow us to make strong developmental statements about the role of this factor. We are unable to make these statements because most

Social Influences on Cognition

117

studies have failed to include multiple age groups. Also, because most studies have involved a single interactive session in which the child makes progress but does not actually reach expert levels of performance, we are unable to trace the changes that underlie a child’s transition from novice to expert. As discussed, this problem is augumented by the fact that many studies have not included pretests or outcome measures. Finally, because high-ability children are often performing at or near ceiling accuracy prior to the interactive session, researchers have difficulty detecting improvements in their task performance. We speculate that because preschool children are more likely to be universal novices, they may be much more dependent on the scaffolding provided by a more expert partner than older children who can at least draw on their general cognitive and social skills to maintain collaboration and to compensate for inadequate task knowledge or to solve problems on their own if a partner is not available. In support of this hypothesis, Azmitia and Bramel (1989) found that expertise and task difficulty manipulations have a stronger effect on 6-year-olds’ than on 9-year-olds’ interaction styles and task performance. Another possibility is that expertise influences not the amount of social facilitation but rather the type of social facilitation. For example, if the child is too inexperienced, a partner’s major contribution may be to keep the child from giving up, but if the child is somewhat more experienced, the social agent may be able to provide more substantive guidance. The contribution of expertise to interactive effects may also vary as a function of whether the child is the recipient or the source of guidance. For example, a recent metanalysis by Cohen, Kulik, and Kulik (1982) showed that regardless of their ability level, tutors tend to benefit more than tutees from interactive problem solving. They argued that this pattern occurred because teaching allows children to organize their representations of problems better. The tutees, however, may not have a sufficiently developed problem representation to assimilate the tutors’ guidance. Thus, they may be haphazardly aggregating information and consequently fail to show improvement even when receiving high-quality instruction from an expert. An extrapolation from this finding is that young children or older novices may fail to transfer their skills beyond interaction because their lack of a sufficiently well-developed problem representation prevents them from assimilating the information provided by others. At this low level of competence, interactive benefits may be confined to maintaining task engagement without necessarily improving task proficiency.

4.

The Relation between Interaction Style and Learning

The issue of the relationship between interaction styles and interactive benefits centers around whether cooperative styles are more likely to produce progress than dominant styles. The majority of studies involving elementary

118

Margarita Azmitia and Marion Rrlmutter

school children have suggested that cooperative styles are more conducive to progress than interactions where one child dominates (Bearison el al., 1986; Doise & Mugny, 1984; Forman & Cazden, 1985; Renshaw & Garton, in press). Researchers have claimed that this pattern occurs because children progress only to the extent that they actively participate in the problem-solving process and that cooperation is more likely to allow both members of the interaction to participate. However, this explanation is challenged by studies of children’s ability to learn from a model (e.g., Botvin & Murray, 1975; Rosenthal & Zimmerman, 1972; Zimmerman & Blom, 1983). They have demonstrated that observational learning can occur without the child actively participating in the task. One problem inherent in the evaluation of the effectiveness of the different interaction styles is that “active participation” nas not been operationalized. For example, a novice who is dominated by the expert during interaction may be participating as actively as possible, given his or her inexperience. Also, it may be that because young children are more inexperienced cognitively and socially, dominant styles may be more prevalent in preschool children than cooperative styles, which require more careful monitoring of the moment-tomoment dynamics of interaction. In this case, the question of which interaction style is more productive would be irrelevant unless one is interested in intervention. Moreover, in situations that preclude interaction (e.g., the child is too inexperienced or the partner exhibits excessive dominance), passive observation may be the only avenue for learning. Age may not be the only relevant variable, however. Ellis and Rogoff (1986), for example, have shown that the nature of the task influences children’s interaction styles. In their study of children’s interactions in two classification tasks, they found that tutors were more dominant in a task resembling home activities than in a task resembling school activities. In addition to exploring the contribution of age and task variables to interactive styles, it may be useful to explore changes in the moment-tomoment dynamics of interaction during the course of a session. For example, although dominant styles may be more beneficial at the initial stages of interaction if the child is a novice, cooperative styles may be more beneficial at later stages when the child’s competence has increased. Such changes in interaction styles are inherent in the concept of proximal development. The social agent scaffolds the child’s performance, gradually allowing her more and more participation until the child is able to take a more active role. Although this change in the dynamics of interaction has been explored in parent-child interactions, it has not been explored widely in peer interactions. The available evidence (e.g., Ellis & Rogoff, 1986) suggests that peers may not be as capable as adults of scaffolding another’s performance.

Social Influences on Cognition

119

5. The Relation between Friendship and harning Throughout this discussion, conclusions about the contribution of peer interaction to cognition have been made primarily from studies in which dyads have been formed randomly, often with the exclusion of partners who are friends. Although such procedures enhance the internal validity of the findings by ensuring initial equivalence between conditions, they may foster underestimation of the power of peer interaction because from an early age, children’s peer interactions occur predominantly with friends. Although few researchers have tested this idea, in three of the four available studies (Hockaday, 1984; J. Nelson & Aboud, 1985; Newcomb & Brady, 1982) friends performed significantly better than nonfriends. Although Newcomb, Brady, and Hartup (1979) did not obtain differences between the performances of friends and nonfriends, their task, building a block tower, may not have been complex enough to tap the skills that make friendship advantageous for collaborative learning. Conversations between friends are marked by greater mutuality and involvement than those between nonfriends (Berndt, 1987; Gottman & Parkhurst, 1980), which may indicate that friends are more tuned to each others’ needs and consequently, more likely to provide effective guidance. Friendship can also enhance learning because friends are more likely than nonfriends to (1) share resources and comply which each others’ requests (La Freniere & Charlesworth, 1987; Newcomb & Brady, 1982), (2) resolve conflicts equitably (Hartup & Laursen, 1987; Krappman & Oswald, 1987), (3) give explanations of their actions (J. Nelson & Aboud, 1985), and (4) try to change each other’s opinion by challenging their partner’s position and championing their own (J. Nelson & Aboud, 1985). Because acquaintances are also more likely than friends to accept each other’s opinion without a challenge (perhaps because they are using the collaborative situation as an opportunity to develop their friendship) (J. Nelson & Aboud, 1985), gains seen during interaction may be less likely to be maintained following the interaction-that is, these gains reflect compliance, not changes in cognitive beliefs. A final advantage of working with a friend may be that because friends have negotiated an interaction style prior to the experimental session, they may be able to devote greater attention to the task than nonfriends, who must divide their cognitive and social resources between the task and managing the social interaction. Data from a number of studies support this hypothesis. First, Goldberg and Maccoby (1965) reported that children who worked with the same partner in a series of problem episodes developed more flexible cooperative skills and greater problem-solving proficiency than children who worked with a different partner each time. Second, Damon (1983) found that over the course of a session, partners became better able to anticipate each

120

Margarita Azmitia and Marion &rimutter

other’s needs and resolve conflicts. Third, Forman and Cazden (1985) found that some dyads needed several sessions to develop stable working styles. Fourth, Doyle, Connolly, and Rivers (1980) found that reciprocity was more common between familiar peers and that familiar peers exhibited more complex play than unfamiliar peers. Finally, Azmitia (in progress) observed that during the first minutes of an interaction, children who were not friends spent most of their time discussing the division of labor and attempting to gain control of the interaction, but that this pattern was not characteristic of friends. In addition, she found that low-ability children who were friends progressed more during interaction than those who were not friends, and that experts in different-ability dyads allowed their novice friends more participation than they allowed a novice stranger. In sum, working with a friend may enhance interactive benefits because friends are aware of each other’s strengths and weaknesses, feel more comfortable seeking and offering aid and expressing disagreement, and have developed a mutually agreeable procedure for dividing labor and resolving conflicts.

6. Summary As was the case for the discussion of parent-child interactions, this discussion of peer interaction has highlighted the potential contribution of social interaction to cognitive development and has raised some qualifications. Again, the generalization that interaction always leads to more progress than solitary activity seems untenable. Variables such as children’s expertise, their developmental level, interaction styles, and social skills need to be considered, as do the types of tasks that are used to explore interaction effects. Moreover, we must pay closer attention to the kinds of social effects that are obtained. As this article shows, the same conclusions apply to the relationship between sibling interactions and cognitive development. The comparison between interactions occurring in the context of relationships and those occurring in the absence of a relationship can help us understand the mechanisms of social facilitation. For example, the expertise of the participants may be less important when the participants are friends because they do not need to establish a working hierarchy and can concentrate on the demands of a problem without wasting time and resources negotiating the social aspects of the interaction. Also, relationships may be more important in early peer interactions because older children’s superior social skills allow them to maintain collaboration with a stranger more easily. However, because very young children do not have a very advanced concept of friendship (cf. Furman, 1982), researchers may find it difficult to identify stable relationships between them. In this case, one might find that pairing friends and nonfriends yields equal levels of performance at the younger ages and

Social Influences on Cognition

121

that as children’s concept of friendship develops, working with a friend becomes more beneficial. Regardless of which hypothesis is correct, exploration of children’s peer relationships would help us differentiate social influences because it would help specify the variables that moderate the magnitude and the type of social facilitation. D. SIBLING INTERACTION

1. Importance Although parent-child and peer interactions have received the most attention, siblings are also important. For example, Cicerelli (1972, 1973, 1975) found that older sisters can be as effective in promoting children’s acquisition of cognitive skills as parents; older brothers are not very effective because they tend to compete, not cooperate, with their siblings. The importance of siblings was further demonstrated by Koester and Johnson (1983), who found that as early as the preschool years, older siblings gave more feedback to their younger sibling than to an unrelated but familiar child. Although Koester and Johnson did not relate these differences to changes in task performance, they may be related. In addition, the special relationship of twin siblings provides perhaps the most extreme example of the ways in which social interaction may influence cognition. For example, Ainslie (1985) suggests that the developmental circumstances of twinship result in characteristic patterns of experiencing oneself.

2. Contributions Siblings may contribute to young children’s cognition for three reasons: First, they usually instantiate the slight cognitive asymmetries needed to induce progress and also serve as socially desirable models. Second, they are more socially skilled than their younger siblings, and thus can compensate for the younger child’s social limitations and maintain the interaction. Third, much like friends, siblings are aware of each other’s competencies and limitations and have developed a method for resolving disagreements. Especially at the younger ages, this feature of the sibling relationship may facilitate interaction.

3. Mediating Factors Although siblings can make an important contribution to a child’s cognitive development, in some situations, their influence will not be helpful. Much like peers, siblings often may be unable to provide the quality of input that can be provided by adults because they lack sufficient knowledge about a problem or lack the skills needed to communicate their view effectively. Also,

122

Margarita Azmitia and Marion Perlmutter

because siblings compete for parents’ attention, having a large number of siblings may actually hinder cognitive growth because parents will have less time for each child (Zajonc & Marcus, 1975). The spacing between siblings also may mediate interaction effects, especially when male siblings are involved. From preschool through adolescence, closely spaced boys display more aggression and competition than widely spaced boys and both closely and widely spaced girls. A corollary to this finding is that girls display more nurturance toward their younger siblings than boys (M. E. Wagner, Schubert, & Schubert, 1985). These differences in interaction patterns may mediate cognitive benefits, For example, Cicerelli (1974) found that younger children are more willing to accept guidance from a sibling who is 4 years older than from a closely spaced sibling. Wide spacing also seems to benefit the development of social and communicative skills (see M. E. Wagner ef al., 1985, for a review), which, in turn, may increase the potential that sibling interaction will facilitate cognitive development.

4. Summary Although siblings may foster each other’s cognitive development, one must keep in mind potential limitations in the benefits accrued from their interactions. We agree with M. E. Wagner ef d ’ s evaluation that due to the dearth of studies on sibling interactions, strong conclusions about the role of siblings in cognitive development cannot be drawn at present. Researchers must also test whether their conclusions apply to siblings in general, or they need to be qualified to accommodate differences in gender, spacing, and birth order. Finally, more studies that examine the moment-to-moment dynamics of sibling interactions are needed. To date, the majority of studies have involved correlating sibling variables, such as numerosity, spacing, gender, and birth order with cognitive variables, such as IQ test performance, academic achievement, and communication skills. We believe that explaining the correlations by identifying the interaction variables that mediate or cause them is as important as demonstrating that particular sibling constellations are associated with particular cognitive abilities. E. SUMMARY

This section highlighted the contribution of different social agents to cognitive development. We argued that even though the parent-child relationship may be more instrumental during the early years, peer relationships may be more instrumental during the elementary school years. Although siblings have not received much attention, they share features with both parents and peers and therefore may prove to be the bridge between them. By studying

*

Social Influences on Cognition

123

their influence more extensively, we might be able to understand the importance of symmetric and asymmetric relationships in cognitive development and may also be able to discern how development influences the importance and the contribution of the different social agents to cognition. Our review showed that even though social interaction may benefit cognition, several factors, such as age, expertise, social competence, and the nature and difficulty of the task will influence the type and magnitude of interaction effects. We may find that in certain situations, some of these factors may actually inhibit interactive benefits and yield better solitary than interactive performance. Our review also showed that executive or metacognitive skillsthat is, skills associated with self-regulation-may be more likely to benefit from interaction. However, as the next section points out, because metacognitive skills depend on a child’s problem-solving competence, the relationship between social facilitation and metacognition may not be as direct as it appears. Although much research has been focused on the contribution of the social agent to cognitive development, we believe that more attention must be paid to the child’s contribution. For example, the increasing importance of peer relations during the elementary school years seems to be not only due to increased opportunities to interact with peers and the emergence of skills needed to maintain interaction, but also due to children beginning to seek their peers (and not their parents or teachers) when they need help (cf. NelsonLe Gall & Gumerman, 1984). Also, if the social agent is not very skilled or is of similar social status (eg., is another child rather than an adult), the learner may take a more active role in ensuring that the interaction meets the learner’s needs. We believe that exploring the child’s contribution to the dynamics of interaction can increase our understanding of social influences on cognition. At a simple level, the child makes the interaction possible. When confronted with a difficult problem, over 97% of preschool and elementary school children indicate that they would seek help from others rather than try to help themselves by persisting in their problem-solving attempt or finding a nonsocial aid (e.g., a book) to guide them (Nelson-Le Gall & Gumerman, 1984). The choice of helper depends on the child’s developmental level. Preschool children generally prefer parents and teachers; elementary school children generally prefer peers. At a more complex level, the child’s cognitive and social skills influence the nature of the other’s guidance. Documenting how these skills influence the interaction may help elucidate the mechanisms of facilitation and determine whether these mechanisms change across development. In the next section, we present a framework that begins to (1) specify the interaction between the child, the social agent, and the task demands, and (2) indicate how this interaction influences the type and magnitude of interaction effects.

124

Margarita Azmitia and Marion Arlmutter

IV. A Framework for Considering Developmental Change in Social Influences on Cognition In the preceding sections, we discussed some of the evidence for the complementary role of parents, siblings, and peers in children’s cognitive development. Although we alluded to possible developmental constraints on social facilitation, we pointed to some limitations in the available research that hinder a determination of whether age, task, or an interaction between these two variables is the relevant variable in age-related differences in the benefits accrued from interactive problem solving. In addition, because most research has started from the premise that social interaction facilitates cognitive development, we have not been overly concerned with finding situations that violated this premise and perhaps, on occasion, we have read stronger support for this premise than is warranted by the data. A.

BENEFITS OF FRAMEWORK

The advantage of conceptualizing social influences within the framework provided is that it allows us to refine the concept of social influences and to hypothesize how social influences may change across development. In contrast with past work, which has assumed that social influence is a unitary construct that exerts its effect on task accuracy, we would like to suggest that looking at different levels of influence and at dependent measures that are not solely indicative of task proficiency may be more useful. For example, this approach may lead to the discovery that the nature of social influence changes across development and that limitations exist in the benefits accrued from collaboration. Also, exploring how task difficulty affects the observed level of social influence can help explain the development of self-regulation. That is, difficult tasks may decrease the likelihood of observing some levels of influence even in older children’s or adults’ interactions. Finally, this framework helps integrate a wide range of research findings. For example, researchers have had difficulty explaining why interactions among young children do not always facilitate learning and why children’s expertise and that of their partners often, but not always, influences the amount of learning that will occur. Researchers also have had difficulty reconciling the different mechanisms of social facilitation because they have assumed that they are incompatible (e.g., imitation precludes conflict). As the next section illustrates, our framework allows us to explain these and other puzzles and to generate future research questions. B. DESCRIPTION OF FRAMEWORK

The framework starts with the assumption that within any particular context, the mechanisms and effects of social interaction depend both on an

Social Influences on Cognition

125

individual’s skill level relative to the task at hand and on the skill level of the social partner relative to the individual of focus. More specifically, we propose four levels of social influence that operate between five performance stages. That is, we assume that subjects’ performance corresponds to their skill level relative to task difficulty and that performance can be characterized as uninvolved, engaged, effective, efficient, or generalized. Increasing the child’s developmental level or reducing the task difficulty are assumed to lead to higher levels of performance. Likewise, providing social input is hypothesized to produce higher stages of performance, although this improvement may depend on the skill of the social partner. Initial improvements in performance can occur in situations where partners have the same skill level; however, more advanced performance gains may require expert input. From this perspective, level of social influence is expected to increase with age if task difficulty is held constant and decrease with task difficulty if developmental level is held constant. The framework is summarized in Table I. As may be seen in the table, the mechanisms of social influence that are hypothesized at Level 1 are most similar to those suggested by social learning theorists (e.g., Bandura, 1977; Dollard & Miller, 1950; Miller & Dollard, 1941). At this stage, an individual’s performance alone may be uninvolved-that is, the child may try to d o something, but his or her lack of understanding of the task may lead him or her to abandon the task or fail to put forth his or her best effort. Social input operates on affect to yield engaged, but not necessarily effective, performance. The other person motivates behavior by reinforcing it. This level 1 social input might colloquially be stated simply as “do something.” Such input should modify performance regardless of whether it comes from another individual of equal skill or from a more expert other. However, this improvement may be confined to maintaining the child’s interest in the task or increasing his or her feeling of competence without necessarily improving his or her accuracy. Although this level of influence is most likely to characterize interactions involving preschool children, particularly when working with unfamiliar peers, we believe it could be characteristic of social influence in all extremely unfamiliar situations. Moreover, if very young children are paired, we believe that even this basic level of social influence may be absent because partners will distract each other from the task (cf. Leuba, 1933). Although the increased engagement that this level of influence can foster may sometimes lead to improved accuracy, this result will not always be achieved because very young children may not be able or willing to conceptualize the task as conceived by the experimenter. Thus, the child may be more engaged, but his or her product may be completely inaccurate when assessed by the experimenter’s criteria. Older children may be able to draw on their more advanced problemsolving repertoire to solve a problem; but if the problem is too difficult, social influence also may be confined predominantly to this low level.

TABLE I Summary of Four-Level Model of Social Influence o n Problem Solving

Level of social influence

Subject’s behavior Alone

6

With another

Level 1: Affect

Uninvolved ..-, Engaged

Level 2: Action

Engaged

Level 3: Strategy

Level 4: Understanding

Examples of relevent measures of subject’s performance

Function of other’s behavior

Other’s behavior

Affect Time on task Number of responses

Motivates behavior

Reinforces

..-, Effective

Diversity of responses Number of correct responses

Varies behavior

Conflicts

Effective

+ Efficient

Percentage of correct responses Time to reach a correct response

Regulates behavior

Directs

Efficient

--t

Transfer performance Verbal justification

Conceptualizes behavior

Generalized

Ex p 1a i ns

Social Influences on Cognition

127

We realize that all of the aforementioned problems (e.g., distraction, inaccurate definition of the task) are not specific to collaborative problem solving; they also occur in solitary problem solving. However, we think they can render collaborative problem solving less advantageous. In cases where they occur, input from an expert peer, older sibling, or adult may be necessary to facilitate problem solving. At Level 2, the mechanisms of social input are most similar to those suggested by Piaget (1968). When an individual’s solitary performance is engaged but not effective (e.g., he or she is not using an appropriate strategy), social input operates on action and leads to effective but not necessarily efficient performance. The other person’s behavior conflicts or challenges the child’s and leads to variation in behavior and thought. At this level, the social input catalyzes change but does not actually provide the substance of it (e.g., the social input provokes a disequilibrium that the child resolves on his or her own). This Level 2 social input might be colloquially stated as “here is another way of doing it.” Such input should improve performance regardless of whether it comes from a matched or more expert other, provided that the child recognizes that there is conflict between his or her view and his or her partner’s view and also recognizes that the task is within a range of his or her skill. That is, the child must have sufficient knowledge about the problem (e.g., be a transitional conserver) to recognize and carry out alternative solutions. Because children are more likely to pay attention to behaviors that they perceive as more advanced than their own (cf. Bandura, 1977), expert input may increase the chances of observing variations in the child’s behavior. However, we admit that we still cannot explain how children come to realize that their partner’s behavior is more advanced, particularly in situations (e.g., conservation tasks) in which the task does not give clear feedback about the adequacy of different solutions. Although this realization may be more likely at transitional stages (as proposed by Piaget), we need to examine collaborative problem-solving interactions more microscopically to see if we can detect the antecedents of this realization. (Hawkins [I9871 suggests that this realizationthat one’s strategy is not as good as one’s partner’s-may be mediated, at least in some situations, by the timing of feedback provided by the partner, but much work still needs to be done to discover the mechanism that leads individuals to reflect upon a problem, to see it in a new light, and consequently, to recognize that one strategy is better than another strategy for attaining the solution.) At Level 3, the mechanisms of social input are most similar to those suggested by Vygotsky (1978). When a person’s solitary performance is effective but not efficient, social input may operate on strategy to yield efficient but not necessarily generalized performance. That is, the child might become capable of selecting the best strategy for a problem, but his or her skills will

I28

Margarita Azmitia and Marion Arlmutter

be context bound because the child does not fully understand the relationship between the problem and the strategy, and between the problem and its variations. The other person’s input directs or regulates the child’s thought or behavior and thereby increases its efficiency. The substance of the other person’s regulation is actually internalized and used to coordinate the child’s behavior. Level 3 social input would be colloquially stated as “do it this better way.” Such input is only expected to improve performance if it comes from another individual who is more expert than the subject, regardless of whether this person is a parent, sibling, or peer. This level of regulation is likely to be largely absent in preschool children’s interactions because they lack the skills to successfully regulate another child’s problem-solving performance or to justify the advantages of their approach. Only when the task is fairly simple and familiar to both children will this level of influence be expected in preschool settings. Finally, at Level 4, mechanisms of social input are those that seem to be assumed by most pedagogical practice. When solitary performance is efficient but not generalized, social input should operate on understanding to yield generalized performance. The other person can provide a conceptualization or explanation of a task or situation. The input at Level 4 might be viewed as metainformation. It would be colloquially stated as “it works because. . . !’ Level 4 social input is only expected to improve performance if it comes from another individual who is more expert than the subject, because only an expert would have the understanding of the problem required to justify strategy selection and adjustments to the problem-solving process and the conviction (cf. Kuhn & Phelps, 1982) to argue their case. Level 4 also would be expected to affect performance only when the child has attained a certain level of task proficiency. This proficiency would include having formed a mental representation of the problem which allows the assimilation of new information in such a way that its significance is recognized. If the child lacks this prerequisite knowledge, he or she merely will be aggregating the information provided by the other and, although he or she may show improvement during the interaction, this improvement would not be expected to continue following the interaction. That is, the child would not fully understand that “it works because.. .” and thus would be unlikely to think of using the strategy in the future. Again, this level of influence will not generally be observed in very young children or in novices because they lack a sufficiently developed representation of the problem at hand. However, if this level of influence operates in early peer relations, we hypothesize that it will be observed in interactions involving older, more expert social agents or in interactions involving friends, or at least involving children with an interaction history. We think that it is important to observe the performance of young children or novices over several interactive sessions to determine whether

Social influences on Cognition

129

their acquisition of task proficiency is accompanied by increasingly higher levels of influence. Figure 1 portrays the hypothesized progression of social influence by showing the quality of performance expected without social input, with matched social input, and with expert social input. As is indicated, social input generally improves performance, although, expecially at the younger ages, the improvement may not be evident by criteria pertaining solely to task accuracy. However, improvement is expected only within a limited range from that attained in solitary activity. Maximal gain may require expert input. The assumption that social input improves performance only within a limited range is similar to the idea of a zone of proximal development, which is central to Vygotsky's theory and is also consistent with numerous formulations that incorporate a notion of optimal discrepancy (e.g., Kagan, 1970). The assumption that expert input may be required for maximal improvements in performance is an extension of previous formulations that generally have been inexplicit on this point. Social input from a similarly skilled partner is expected to facilitate performance only through Level 1 and Level 2. Although a similarly skilled other person may reinforce (Level l), conflict (Level 2), direct (Level 3), or explain (Level 4) behavior, their directions and explanations are not likely to be very useful. However, reinforcement and conflict, even from

SOCIAL INFLUENCE

SUBJECT'S SKILL RELATIVE TO TASK D I F F I C U L T Y

LEVEL 1 LEVEL 1

~

AFFECT

~

ACTION

~

UNDERSTANDING

PERFORMANCE

+

UNINVOLVED

+ - --

-___-________-__ _______-_-

Developmsnr LEVEL 5

-

- _ - - - _ _ - _ _ - - _ - -_- _ r 7 - - -I-

Development

LEVEL 4

-

EXPERT SOCIAL INPUT

_ _ _ _ - _ _ _ _ _ _ _ _ _ _ _ - _ - - _E N -G A-G E_D - - - -

Zone of Proximal

LEVEL 4

MATCHED SOCIAL INPUT

_______________

Zone of Proximal Development

LEVEL 3

L E V E L 3 STRATEGY

-

Zone of Proximal De"e1opnlent

LEVEL 2

LEVEL 2

NO SOCIAL INPUT

EFFECTIVE

EFFICIENT

ALIZED - _ - _ - _ _ _ _ _ - _ _ - _ _ - _ _ _+_G-E N_E R_ -__

Fig. 1. Hypothesized progression of social influence as a function of the subjecti skill and the type of social input.

130

Matgarita Azmitia and Marion Arlmutter

a similarly skilled other, should reinforce or vary behavior and therefore lead to potentially higher stages of performance. As suggested in Table I, different measures of subjects’ performance are most useful for inferring each level of social influence. Level 1 influence is assumed to operate on affect and to lead to more engaged behavior. It should increase positive affect, increase time on task, and increase number of responses. Level 2 influence is assumed to operate on action and to lead to more effective behavior. It should increase the diversity of responses and the number of correct responses. Level 3 influence is assumed to operate on strategy and contribute to more efficient behavior. It should increase the percentage of correct responses and decrease the time taken to make a correct response. Finally, Level 4 influence is assumed to operate on understanding and to lead to more generalized behavior. It should enhance transfer and lead to more adequate verbal justification. Up to this point, we have been fairly inexplicit about how task variables influence the interaction between developmental level and type of social influence. We would like to offer some examples of how task variables may influence this interaction. Most studies of social influence on cognition have focused on social interaction during problem solving. However, especially for younger children, problem-solving contexts may be too limiting. For example, Brownell (1982) observed that toddlers exhibited more sophisticated interactive skills in a free-play situation than in a problem-solving situation. She speculated that this difference stems from differences in the demands posed by such settings. Play contexts allow children to “go with the flow” and alter their goals at will. In contrast, more formal problems usually have only one solution, and thus children must subordinate their interactions to a single goal. For reasons discussed here previously, this restriction may prove quite difficult for young children. However, the constraint of having to subordinate the interaction to one goal may not be the only variable mediating the lack of sophistication exhibited by young children’s problem-solving interactions. Garvey (1987) reported that young children strive to maintain the structure of pretend play, thus demonstrating that they can subordinate their interactions to attain a goal of their choice-that is, to maintain and elaborate the pretendplay structure. Thus, we must also consider how appealing our tasks are to children, inasmuch as an appealing task will motivate them to display more sophisticated interactions aimed at solving the task efficiently. Finally, certain tasks may be more appropriate for certain ages. For example, construction tasks appear to be one of the few that reveal the cognitive benefits of collaboration during the preschool years. These tasks generally do not require explanations or discussions; rather, children can learn from observing their partner (see, for example, Azmitia, 1987; Morrison & Kuhn, 1983). In contrast, during the elementary school years, construction tasks usually d o not yield collaborative benefits, but tasks (e.g., math, computer program-

Social Influences on Cognition

131

ming) that require explanation and discussion reveal the cognitive gains accrued from collaboration (see, for example, Hawkins et al., 1984; Phelps & Damon, 1987). Manipulating task variables also can help identify developmental sequences in interactive problem solving. Although most studies have been focused on a single age-group or on a task that is relevant for only a very narrow age range, selecting a task that is meaningful for a broad developmental range may provide valuable information. For example, this approach may indicate whether social influences merely accelerate development or actually alter its course. This approach also may allow us to specify more accurately how a child’s general developmental level mediates interactive problem solving. For example, we have assumed that as children get older, they acquire general cognitive and social heuristics that allow them to collaborate in a wide range of problems. However, because comparisons across age-groups often confound age and task, we have had difficulty isolating the effects of these variables. Developmental comparisons on the same task may reveal that although differences in general problem solving and social heuristics contribute to age-related differences in interactive problem solving, differences in children’s tendency to apply these heuristics to specific problems also play a role. Moreover, developmental comparisons may reveal that interaction benefits are more likely at certain points during the development of a skill. C . SUMMARY

We have presented a discussion of how social input may have different types of influence on cognition. Briefly, we proposed that social influence is better conceptualized as operating within four levels, the lowest being affect and the highest being metacognition. The lack of relevant data prevents us from making strong developmental statements, but we think that at the younger ages, only parents and older siblings may have the expertise needed to influence efficiency (i.e., selection of the best strategy) and understanding (i.e., this strategy is best for this reason). However, if the task is quite simple, in some instances (e.g., when the children are friends), even interactions among young children may contain elements of these higher levels of influence. Although older children’s interactions with peers are likely to span the four levels of influence, if the task is too difficult, or at least one child is not an expert, the two highest levels of influence also are likely to be absent.

V. Additional Theoretical and Methodological Issues In this final section, we discuss some additional theoretical issues and some of the data that we believe must be gathered in order to increase the understanding of social influences on cognition.

132

Matgarita Azmitia and Marion firlmutter A. THEORETICAL ISSUES

Thus far, we have emphasized general developmental sequences and how these sequences may mediate the type of social influence on cognition and the magnitude of social facilitation. Although we argued that individual differences in children’s developmental levels can affect patterns of collaboration, we would like to offer some additional examples of how individual differences may influence patterns of social facilitation. Only a handful of studies have been addressed to the contribution of individual differences to social interaction. Nevertheless, these data are very promising. Earlier, we mentioned that several researchers (e.g., Brown & Ferrara, 1985; Feurstein et al., 1981) have tried to relate individual differences in zones of proximal development to academic achievement. In addition, some attempt has been made to assess the contribution of personality factors to collaborative problem solving. For example, Webb (1980) found that extroverted junior high school students benefited more than introverted students did from collaborative problem solving. Extroverts were less likely to be ignored and more likely to have their questions answered than introverts. The fact that getting questions answered was the best predictor of learning suggests that this difference may explain why extroverts benefited more from collaboration. Other individual difference variables may also contribute to interactive problem solving. Independent of age, interaction history, and ability, some dyads are able to work together quite efficiently, and others are unable to do so. In our own research, we have been impressed by the great individual differences in children’s responses to interactive instructions. For example, some children immmediately begin to work in parallel, while others devote a few minutes to discuss the distribution of labor and then adopt parallel styles. Still others adopt a dominant style where one does most of the work and the other observes and makes suggestions that are largely ignored by the dominant partner. However, not all dominant styles instantiate a worker and an observer; some denote interactions in which one child becomes the tutor and the other the learner. Finally, some interactions instantiate true cooperation: Children discuss the problem, take turns, and support each others’ work by giving and accepting suggestions and by providing emotional support. In agreement with our own informal observations, teachers often remark that they are curious to see how Xwill do in our task because he or she is a loner or too bossy, or that they are sure that Y will do beautifully because he or she works well with others. Systematic study of what accounts for these differences may reveal that some children learn best through solitary work and that others learn best through collaboration. These systematic observations may also help us understand why dyads select particular interaction styles.

Social Influences on Cognition

133

Individual differences in cognitive styles may also mediate the type and magnitude of social facilitation. For example, impulsive children may benefit more from interactive problem solving than reflexive children because they are forced to slow down and consider their partner’s perspective before committing themselves to a solution. Although reflexive children may also benefit from collaboration, the types of benefits may be different from those accrued by impulsive children because reflexive children already know the value of considering alternatives and evaluating their work before committing themselves to an answer. B. METHODOLOGlCAL ISSUES

We have already discussed the value of considering (1) the relationship between social and cognitive skills, (2) the child’s cues in relation to the partner’s guidance, (3) the task demands, and (4) the level of the partner’s expertise when attempting to predict the type and magnitude of social facilitation. Here, we would like to highlight the potential contribution of observational and longitudinal data.

1. The Importance of Collecting Observational Data Most data about social influences on cognitive development have come from laboratory studies. Unfortunately, the results may have been influenced by the constraints of the laboratory. For example, although we can observe interactions in the natural setting without the participants knowing that they are being observed, this inobtrusiveness usually is impossible in the laboratory (unless a long familiarization time is involved during which children come to ignore the observer, as is the case in some observations of interactions in preschool classrooms (H. W. Reese, personal communication, August 17, 1987). If the observer is not inobtrusive, however, his or her presence may alter the nature of interactions. In support of this hypothesis, Graves and Glick (1978) found that when they thought they were being observed, mothers doubled their amount of speech, directed more of their speech to the child, and seemed to be actively engaged in teaching their children new skills or getting them to display skills that they had already acquired. Thus, laboratory settings can lead to overestimations of the amount of support that parents usually give their children during interactive problem solving. Our informal observations of preschool children’s interactions in the laboratory also suggest that being observed can influence the pattern of interaction. For example, relative to interactions in the classroom, their verbal output seems to decrease and the length of the interaction to increase. Children seem to maintain each other’s task engagement by reminding each other that the experimenter told them to work together and to finish the task.

134

Margarita Azmitia and Marion &rimutter

It would be important to know how individuals maintain each other’s task engagement when competing activities are available, and when the presence of the experimenter does not constrain their participation in other activities. We might find, for example, that given the greater number of interruptions and the more informal structure of everyday settings, interactions in natural settings are characterized by brief, intense exchanges rather than by the smooth, lengthier exchanges that are usually observed in the laboratory. An additional advantage of conducting observations in the everyday environment is that this approach may indicate which tasks or skills tend to lend themselves to interaction. Too often, the contribution of social interaction to the development of cognitive skill focuses on cognitive skills that have been identified as important only in laboratory tasks. Moreover, when investigation of social influence is carried out in the laboratory, we can only determine whether interaction can facilitate the acquisition of a skill, not whether the skill is usually acquired through interaction. Although we believe that observational studies can provide important information, we acknowledge that observations in unstructured situations often will be unfeasible because the criteria1 behaviors are too infrequent or too unpredictable. However, the frequency and predictability of a behavior can provide information about the importance of the behavior in the child’s overall repertoire, Also, observational studies can give perspective to patterns that emerge in the laboratory. For example, the finding that preschool children exhibit more dominance than elementary school children during interactive problem solving in the laboratory appears to stem from the fact that dominant styles predominate in their everyday interactions (cf. Strayer & Trudel, 1984).

2. The Importance of Collecting Longitudinal Data In addition to collecting observational data, we must also collect longitudinal data because such data provide the only basis to establish causeeffect relationships between a child’s cognitive and social skills and the observed level of social influence. For example, longitudinal research would permit evaluation of the hypothesis that low levels of facilitation, such as task engagement and enjoyment, set the stage for higher levels of influence, such as strategy selection and understanding. In addition, longitudinal data can aid in the identification of factors that are responsible for the salience of different social agents across development. Longitudinal data would also allow us to track changes in interaction styles and to map these changes onto changes in cognitive performance. For example, Selman and Yeates (1987), who longitudinally studied interactions between preadolescents found that during the initial weeks of acquaintanceship, partners devoted most of their energy to competing and jockeying for status, and

Social Influences on Cognition

135

only after several weeks had gone by did they start building the shared understanding (which comes about through shared experiences) that may be necessary for reaping the cognitive benefits of collaboration. A similar pattern was reported by Forman and Cazden (1985), who studied dyads’ performance in the chemical solutions problem over an 11-week period. Specifically, cooperative interactions in which children monitored each others’ performance and coordinated their plans were rare during the initial sessions, but during the final sessions, these interactions constituted the bulk of interactions for two of the three dyads. As mentioned, mutuality and shared understanding may underlie the differences between the learning accrued from problemsolving interactions between children and their parents, children and their friends, and children and unfamiliar peers. Longitudinal data also can be used to assess the long-term consequences of mediated learning. For example, how does solving a problem interactively influence the child’s subsequent performance on related and unrelated problems? Perhaps an enjoyable social experience increases the likelihood that children will seek interactive contexts in the future. Also, some types of social facilitation may not be evident in immediate posttests but will emerge over time as individuals consolidate their strategies and gain understanding about the relation between certain strategies and certain problems. Although longitudinal studies usually represent a great expenditure of effort and are often difficult to carry out, we think that their potential contribution to understanding cognitive development, particularly social influences on cognitive development, makes them well worth the effort.

VI. Conclusions Two goals of this article are (1) to present a developmental framework to evaluate social influences on young children’s cognition and (2) to offer some suggestions for future research. The exploration of sociocognitive interactions has had a long theoretical history, but only since the late 1970s has it become one of the leading problems studied by developmental psychologists. Our review of the literature suggested that interactions with adults, peers, and siblings may facilitate young children’s acquisition of cognitive skills. However, the assumption that all cognitive skills originate in social interaction appears untenable. In some situations, solitary work is equally or more conducive to learning. Thus, an important task for the future is to identify which skills are most likely to benefit from social interaction and whether at different points in development some social agents are more important for learning. In particular, we argued that even though researchers have assumed that peers are less important than adults during the early years, research that has yielded

136

Margarita Azrnitia and Marion Perlmutter

evaluations of peer influences may have had limitations that led to underestimations of the importance of peers. Although much research has suggested the importance of the social context in development, this work has neither been integrated nor considered with respect to possible developmental constraints. To begin to alleviate these limitations, we have offered a preliminary framework of the way the nature of social influences may change with age and task complexity. We also suggested that we will have to expand the range of measures that we use to study social influences. In particular, we need to stop relying on task accuracy as the primary measure of social facilitation. We also offered some directions for future research. In particular, we argued that the range of tasks, contexts, and regulators of interaction need to be expanded. Especially at younger ages, we may need to move out of the laboratory and the previous focus on laboratory-conceived problem-solving tasks to examine interaction in more informal contexts and more natural tasks. Although past research has tended to focus on how social agents transmit information to the child, we think it essential now to consider also how the child contributes to his or her own development. That is, we need to focus on interpersonal and intrapersonal development and on the interaction between them. Moreover, researchers need to focus more on individual differences in cognitive and interactive styles. Finally, we need to begin to design studies that will establish causal connections between social interaction and cognitive development. Specifically, researchers should measure children’s competence prior to and following the interaction and should empirically manipulate, inasmuch as possible, the variables that are hypothesized to be responsible for the cognitive benefits accrued from interaction. Longitudinal data are likely to provide valuable insights about the mechanisms that promote the internalization of the other’s guidance and the ways that social interaction influences the child’s cognitive competence. ACKNOWLEDGMENTS We are grateful to several individuals whose contributions greatly improved the quality of this article. In particular, Stephanie Behrend, Doug Behrend, Hayne Reese, and Barbara Rogoff provided insightful comments on earlier versions of the manuscript, Mildred Alvarez and Bill Hartup helped orient us to the literature on social development, and Susan Kemper helped refine our grasp on cognitive issues. Also, the Brookdale Foundation’s National Fellowship Award provided support for Marion Perlmutter during the writing of this article.

REFERENCES Ainslie, R. C. (1985). The psychology of twinship. Lincoln: University of Nebraska Press. Allen, V. L. (Ed.) (1973). Children as teachers. New York: Academic Press.

Social Influences on Cognition

137

Ames, G. J., & Murray, F. B. (1982). When 2 wrongs make a right: Promoting cognitive change by social conflict. Developmental Psychology, 18, 894-897. Ashworth, P. 0. (1979). Social interaction and consciousness. New York: Wiley. Azmitia, M. (1987, April). Expertise as a moderator of social influence on children’s cognition. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Azmitia, M. (1988). Peer interaction and problem solving: When are two heads better than one? Child Development, 59, 87-96. Azmitia, M., & Bramel, M. (1989). “You’vegot the looks but I’ve got the brains.” The influence of age, expertise, and task dirfinrlty on children’sco/labomtiveproblem solving. Unpublished manuscript, Florida International University. Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall. Bearison, D. J. (1982). New directions in studies of social interaction and cognitive growth. In F. Serafica (Ed.), Social-cognitive development in context (pp. 199-221). New York: Guilford. Bearison, D, J., & Cassel, T. (1975). Cognitive decentration and social codes: communicative effectivenessand coordination of perspectives in young children. Developmental Psycholou, 11, 29-36. Bearison, D. J., Magzamen, S., & Filardo, E. K. (1986). Socio-cognitive conflict and cognitive growth in young children. Merrill-Palmer Quarterly, 32, 51-72. Bee, H. L., Van Egeren, L. F., Streissguth, A. P., Nyman, B. A., & Fleckie, M. S. (1969). Social class differences in maternal teaching strategies and speech patterns. Developmental Psychology, 1, 726-734. Behrend, D. A., Rosengren, K. S., & Perlmutter, M. (1986). Privatespeech in preschool children’s problem solving alone and with their mother. Unpublished manuscript, University of Michigan. Berndt, T. J. (1987). The distinctive features of conversations between friends: Theories, research, and implications for socio-moral development. In W. M. Kurtines & J. L. Gewritz (Eds.), Moral development through social interaction (pp. 281-300). New York: Wiley. Botvin, G: J., & Murray, F. B. (1975). The efficacy of peer modeling and social conflict in the on of conservation. Child Development, 46,796-799. Brenner, J., & Mueller, E. (1982). Shared meaning in boy toddlers’ peer relations. Child Development, 53, 380-391. Brown, A. L., Bransford, J. D., Ferrara, R. A., & Campione, J. C. (1983). Learning, remembering, and understanding. In P. H. Mussen (Ed.), Handbook of child psychology: Vol. I. Cognitive development. J. H. FlaveU & E. M. Markman, Eds. @p. 77-166). New York: Wiley. Brown, A. L., & Ferrara, R. A. (1985). Diagnosing zones of proximal development. In J. V. Wertsch (Ed.), Culture, communication and cognitiotr Vygotskianperspectives(pp. 273-305). London & New York: Cambridge University Press. Brownell, C. A. (1982). Peer interaction among toddler-age children: Effects of age and social context on interactional competence and behavioral roles. Unpublished doctoral dissertation, University of Minnesota, Minneapolis. Brownell, C. A., & Brown, E. (1985, April). Toddlerpeer interaction in relationship to cognitive development. Paper presented at the meeting of the Society for Research in Child Development, Toronto, Canada. Bruner, J. S. (1977). Early social interaction and language acquisition. In H. R. Schaffer (Ed.), Studies in mother-infant interaction (pp. 271-289). London: Academic Press. Bruner, J. S. (1985). Vygotsky: A historical and conceptual perspective. In J. V. Wertsch (Ed.), Culture, communication and cognition: Vygotskianperspectives (pp. 21-34). London & New York: Cambridge University Press. Campione, J. C., Brown, A. L., Ferrara, R. A., & Bryant, N. R. (1984). The zone of proximal development: Implications for individual differences and learning. In B. Rogoff & J. V. Wertsch (Eds.), Children’s learning in the zone of proximal development (pp. 77-92). San Francisco, CA: Jossey-Bass.

138

Margarita Azmitia and Marion Perlmutter

Cantor, G. N. (1983). Conflict, learning, and Piaget: Comments on Zimmerman and Blom’s “Toward an empirical test of the role of cognitive conflict and learning.” Developmental Review, 3, 39-53. Cavanaugh, J. C., & Perlmutter, M. (1982). Metamemory: A critical evaluation. Child Development, 53, 11-28. Cazden, C. B., Cox, M., Dickinson, D., Steinberg, Z., & Stone, C. (1979). You all gonna hafta listen: Peer teaching in a primary classroom. In W. A. Collins (Ed.), Minnesota Symposia on Child Psychology (Vol. 12, pp. 183-231). Hillsdale, NJ: Erlbaum. Cicerelli, V. G . (1972). The effect of the sibling relationship on concept learning of young children taught by child teachers. Child Development, 43, 282-287. Cicerelli, V. G . (1973). Effects of sibling structure and interaction on children’s categorization style. Developmental Psychology, 9, 132-139. Cicerelli, V. G. (1974). Relation of sibling structure and interaction to younger siblings’ style. Journal of Genetic Psychology, 125, 37-49. Cicerelli, V. G. (1975). Effect of mother and older sibling on the problem-solving behavior on the younger child. Developmental Psychology, 11, 749-756. Cohen, P. A., Kulik, J. A., & Kulik, C. D. (1982). Educational outcomes of tutoring: A metaanalysis. American Educational Research Journal, 19, 237-248. Cole, M. (1985). The zone of proximal development: Where culture and cognition create each other. In J. V. Wertsch (Ed.), Culture, communication, and cognition: Vygotskian perspectives (pp. 146-161). London & New York: Cambridge University Press. Condon, W. S., & Sander, L. W. (1974). Neonate movement is synchronized with adult speech. Science, 183, 99-101. Cooper, C. R. (1980). Development of collaborative problem solving among preschool children. Developmental Psychology, 16, 433-444. Cooper, C. R., Ayers-Lopez, S., & Marquis, A. (1982). Children’s discourse during peer learning in experimental and naturalistic situations. Discourse Processes, 5, 191-197. Dale, P. S. (1986, March). Language developmenr as outcome and mediator in early intervention. Paper presented at the Child Language Proseminar, University of Kansas, Lawrence. Damon, W. (1983). The nature of social-cognitive change in the developing child. In W. F. Overton (Ed.), The relationship between social and cognitive development (pp. 103-139). Hillsdale, NJ: Erlbaum. Davis, A., & Lange, G. (1973). Parent-child communication and the development of categorization styles in preschool children. Child Development, 44, 624-629. DeLoache, J. S. (1983, April). Joint picturebook reading as memory training. Paper presented at the meeting of the Society for Research in Child Development, Detroit, MI. DeLoache, J. S., & Plaetzer, B. (1985, April). Tea for two: Joint mother-child symbolic play. Paper presented at the meeting of the Society for Research in Child Development, Toronto, Canada. Doise, W., & Mugny, G. (1979). Individual and collective conflicts of centrations and cognitive development. European Journal of Social Psychology, 9, 105-109. Doise, W., & Mugny, G. (1984). The social development of the intellect. Oxford, Pergamon. Doise, W., Mugny, G., & Perret-Clermont, A.-N. (1975). Social interaction and the development of cognitive operations. European Journal of Social Psychology, 5, 367-383. Doise, W., Mugny, G., & Perret-Clermont, A.-N. (1976). Social interaction and the development of logical operations. European Journal of Social Psychology, 6, 245-247. Dollard, J., & Miller, N. E. (1950). Personality and psychology. New York: McGraw-Hill. Donaldson, M. (1978). Children’s minds. New York: W. W. Norton. Doyle, A., Connolly, J., & Rivers, L. (1980). The effects of playmate familiarity on the social interactions of young children. Child Development, 51, 217-223. Eder, R. A,, Parks, C., Todd, C. M., Robb, M., & Perlmutter, M. (1986). Parental effecfs on

Social Influences on Cognition

139

preschool children’s memory. Unpublished manuscript, Southern Methodist University, Dallas, TX. Ellis, S., & Rogoff, B. (1982). The strategies and efficacy of child versus adult teachers. Child Development, 53, 730-735. Ellis, S., & Rogoff, B. (1986). Problem solving in children’s management of instruction. In E. C. Mueller & C. R. Cooper (Eds.), Process and outcome in peer relationships (pp. 301-325). Orlando, FL: Academic Press. Feurstein, R., Miller, R., Rand, Y., & Jensen, M. R. (1981). Can evolving techniques better measure cognitive change? Journal of Special Education, 15, 201-219. Field, D. (1981). Can preschoolers really learn to conserve? Child Development, 52, 326-334. Fischer, K. W. (1980). A theory of cognitive development: The control and construction of hierarchies of skills. Psychological Review, 87, 477-531. Flavell, J. H., Botkin, P. T., Fry, C. L., Wright, J. W., & Jarvis, P. E. (1968). The development of role taking and communication skills. New York: Wiley. Forman, E. A., & Cazden, C. B. (1985). Exploring Vygotskian perspectives in education: The cognitive value of peer interaction. In J. V. Wertsch (Ed.), Culture, communication and cognition: Vygotskianperspecfives(pp. 323-347). London & New York: Cambridge University Press. Furman, W. (1982).Children’s friendships. In T. Field, G. Finley, A. Huston, H. Quay, & L. Troll (Eds.), Review of human development (pp. 327-342). New York: Wiley. Garvey, C. (1987, April). Creation and avoidance of conflict in preschool children’splay. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Gauvain, M., & Rogoff, B. (1985, April). The development ofplanningskills by individuals and dyads. Paper presented at the meeting of the Society for Research in Child Development, Toronto, Canada. Gelman, R. (1978). Cognitive development. Annual Review of Psychology, 29, 297-331. Gewirtz, J. L. (1969). Mechanisms of social learning: Some roles of stimulation and behavior. In D. A. G o s h (Ed.), Handbook of socialization theory and research (p. 57-212). New York: Atherton. Gick, M. L., & Holyoak, K. J. (1980). Analogical problem solving. Cognitive Psychology, 12, 306-355. Gick, M. L., & Holyoak, K. J. (1983). Schema induction and analogical transfer. Cognitive Psychology, 15, 1-38. Glachan, M., & Light, P.(1982). Peer interaction and learning: Can two wrongs make a right? In G. Butterworth & P. Light (Eds.), Social cognition: Studies in the devefopment of understanding (pp. 238-262). Chicago, IL: University of Chicago Press. Goldberg, M. H., & Maccoby, E. E. (1965). Children’s acquisition of skill in performing a group task under 2 conditions of group formation. Journal of Personalifyand Social Psychology, 2, 898-902. Goncu, A,, & Rogoff, B. (1987, April). Nature of adult guidance as a moderator of social influence on children’s cognition. Paper presented at the meeting of the Society for Research in Chld Development, Baltimore, MD. Goodsitt, J., Grady-Reitan, J., & Perlmutter, M. (in press). Interaction between mothers and preschool children when reading novel and familiar books. International Journal of Behavioral Development. Gottman, J. M., & Parkhurst, J. T. (1980). A development theory of friendship and acquaintanceship processess. In W. A. Collins (Ed.), Minnesota Symposia on Child Psychology (Vol. 13, pp. 197-254). Hillsdale, NJ: Erlbaum. Graves, Z. R., & Click, J. (1978). The effect of context on mother-child interaction: A progress report. Quarterly Newsletter of the Institutefor Compamtive Human Development, 2,41-46. Greenfield, P., & Lave, J. (1982). Cognitive aspects of informal education. In B. A. Wagner &

140

Margarita Azmitia and Marion Perlmutter

H. W. Stevenson (Eds.), Culturalperspectives on child development (pp. 181-207). San Francisco, CA: Jossey-Bass. Hartup, W. W. (1985). Relationships and their significance in cognitive development. In R. A. Hinde, A.-N. Perret-Clermont, & J. Stevenson-Hinde (Eds.), Relationships and cognitive development (pp. 66-82). London & New York: Oxford University Press (Clarendon). Hartup, W. W., & Laursen, B. (1987, April). Friendship and conflict: Synergies in child development. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Hawkins, J. (1987, April). Co//abomtionand disent. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Hawkins, J., Homolski, M., & Heide, P. (1984). Pairedproblem solving in a computer context (Tech. Rep. 33). New York: Bank Street College of Education. Hess, R. D., & Shipman, V. C. (1967). Cognitive elements of maternal behavior. In J. Hill (Ed.), Minnesota symposia on childpsychology (Vol. I, pp. 57-81). Minneapolis: University of Minnesota Press. Hockaday, F. (1984). Collaborativelearning with young children. Educational Studies, 10,237-242. Jaffe, J., Stern, D. N., & Perry, J. C. (1973). “Conversational” coupling of gaze behavior in prelinguistic human development. Journal of Psycholinguistic Research, 2, 321-329. Johnson, D. W., Johnson, R. T., & Roy, P. (1984). Circles of learning. Washington, DC: Association for Supervision and Curriculum Development. Jones, C. P., & Adamson, L. B. (1985, April). Lunguage use in mother-child and mother-childsibling interactions. Paper presented at the meeting of the Society for Research in Child Development, Toronto, Canada. Kagan, J. (1970). The determinants of attention in the infant. American Scientist, 58, 298-306. Killen, M. (1987, April). Contextualfeatures of peer interaction. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Koester, L. S., & Johnson, J. (1983, March). Children’s instructional strategies: A comparison of sibling and peer tutoring. Paper presented at the meeting of the Society for Research in Child Development, Detroit, MI. Kohlberg, L., Yaeger, J., & Hjertholm, E. (1968). Private speech: Four studies and a review of theories. Child Development, 39, 671-690. Kontos, S. (1983). Adult-child interaction and the origins of metacognition. Journal of Educational Research, 77, 43-54. Krappman, L., & Oswald, H. (1987, April). Negotiation strategies in peer cortflicts: A follow-up study in natural settings. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Kuhn, D. (1972). Mechanisms of change in the development of cognitive structures. Child Development, 43 833-844. Kuhn, D., & Phelps, E. (1979). A methodology for observing development of a formal reasoning strategy. In D. Kuhn (Ed.), Intellectual development beyond childhood (pp. 45-58). San Francisco, CA: Jossey-Bass. Kuhn, D., & Phelps, E. (1982). The development of problem solving strategies. Advances in Child Development and Behavior, 17, 2-44. Laboratory of Comparative Human Cognition (1983). Culture and cognitive development. In P. H. Mussen & W. Kessen (Eds.), Handbook of childpsychology: Vol. 1. History, fheory, and methods (pp. 295-356). New York: Wiley. L a Freniere, P. J., & Charlesworth, W. R. (1987). Effects of friendship and dominance status on preschoolers’ resource utilization. International Journal of Behavioral Development, 10, 345-358. Laosa, L. M. (1980) Maternal teaching in Chicano and Anglo-American families: The influence of culture and education on maternal behavior. Child Development, 51, 759-765.

Social Infuences on Cognition

141

Laosa, L. M. (1982). School, occupation, culture, and family: The impact of parental schooling on the parent-child relationship. Journal of Educational Psychology, 14, 791-827. Leuba, C. (1933). An experimental study of rivalry in young children. Journal of Comparative Psychology, 16, 367-378. Luria, A. R. (1976). Cognitive development. Cambridge, MA: Harvard University Press. McNamee, 0. D., McLane, J. B., Cooper, P., & Kerwin, S. (1984). Final report on the origins of literacy during the preschool years (Tech. Rep.). Chicago, IL: Erickson Institute. Mead, G. H. (1934). Mind, se& and society. Chicago, IL: University of Chicago Press. Miller, N. E., & Dollard, J. (1941). Social learning and imitation. New Haven, CT: Yale University Press. Morrison, H., & Kuhn, D. (1983). Cognitive aspects of preschooler’s imitation in a play situation. Child Development, 54, 1054-1063. Moss, E. B. (1984). Parent-child interaction and the development of metacognition. Unpublished manuscript, Universite de Quebec ti Montreal. Mowrer, 0. H. (1960). Learning theory and behavior. New York: Wiley. Mueller, E., & Brenner, J. (1977). The origins of social skills and interaction among toddlers. Child Development, 48, 854-856. Mugny, G., & Doise, W. (1978). Socio-cognitive conflict and structure of individual and collective performances. European Journal of Social Psychology, 8 181-192. Murray, F. B. (1981). The conservation paradigm: Conservation of conservation research. In D. Brodinsky, I. Siegel, & R. Golinkoff (Eds.), New directions in Piagetian theory and research (pp. 143-176). Hillsdale, NJ: Erlbaum. Murray, F. B. (1982). Teaching through social conflict. Contemporary Educational Psychology,

I, 257-271. Murray, F. B. (1983). Equillibration as cognitive conflict. Developmental Review, 3, 54-61. Murray, J. P. (1974). Social learning and cognitive development: Modeling effects on children’s understanding of conservation. British Journal of Psychology, 65, 151-169. Nelson, J., & Aboud, F. (1985). The resolution of social conflict between friends. Child Development, 56, 1009-1017. Nelson, K. (1981). Individual differences in language development: Implications for development and language. Developmental Psychology, 16, 170-187. Nelson-Le Gall, S. A., & Gumerman, R. A. (1984). Children’s perceptions of helpers and helper motivation. Journal of Applied Developmental Psychology, 5, 1-12. Newcomb, A. F., & Brady, J. E. (1982). Mutuality in boys’ friendship relations. Child Development, 53, 392-395. Newcomb, A. F., Brady, J. E., & Hartup, W. W. (1979). Friendship and incentive conditions as determinants of children’s task oriented social behavior. Child Development, 50, 878-881. Ninio, A. (1983). Joint book reading as a multiple vocabulary acquisition device. Developmental Psychology, 19, 445-451. Ninio, A., & Snow, C. E. (1985, April). The contract of literacy: What children learn from reading books. Paper presented at the meeting of the Society for Research in Child Development, Toronto, Canada. Paris, S., & Lindauer, B. (1982). The development of cognitive skills during childhood. In B. Wollman (Ed.), Handbook of developmentalpsychologv @p. 333-349). Englewood Cliffs, NJ: Prentice-Hall. Perlmutter, M. (1980). Children’s memory. San Francisco, CA: Jossey-Bass. Perlmutter, M., Behrend, S., Kuq F., & Muller, A. (in press). Social influence on children’s problem solving at a computer. Developmental Psychology. Perret-Clermont, A.-N. (1980). Social interaction and cognitive development in children. London: Academic Press. Perret-Clermont, A.-N., & Brossard, A. (1985). On the interdigitation of social and cognitive pro-

I42

Margarita Azmitia and Marion Perlmutter

cesses. In R. A. Hinde, A.-N. Perret-Clermont, & J. Stevenson-Hinde(Eds.), Relationships and CognitiveLhelopmenf(pp. 3W327). h n d o n & New York Oxford University Press (Clarendon). Phelps, E., & Damon, W. (1987, April). Peer collaboration andproblem solving. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Piaget, J. (1932). The moral judgement of the child. Glencoe, IL: Free Press. Piaget, J. (1954). The construction of reality in the child. New York: Basic Books. Piaget, J. (1968). Six psychological studies. New York: Vintage Press. Piaget, J. (1977). The equilibration of cognitive structures. Chicago, IL: University of Chicago Press. Pratt, W. W., Scribner, S., & Cole, M. (1977). Children as teachers: Developmental studies of instructional communication. Child Development, 48, 1475-1481. Radin, N. (1982). The unique contribution of parents to child rearing: The preschool years. In S. G . Moore & C. R. Cooper (Eds.), The young child, reviews of psychology (Vol. 3 , pp. 57-78). Washington, DC: National Association for the Education of Young Children. Ratner, H. H. (1980). The role of social context in memory development. In M. Perlmutter, (Ed.), Children’s memory (pp. 49-68). San Francisco, CA: Jossey-Bass. Ratner, H. H. (1984). Memory demands and the development of young children’s memory. Child Developmeni, 55, 2173-2191. Reeve, R. A. (1987, April). The functional significance of paren fa1scaffolding as a moderator of social influence on children’s cognition. Paper presented at the meeting of the Society for Research in Child Development Baltimore, MD. Renninger, K. A., & Morgan, A. (1986, May). Other and play object as value in social understanding. Paper presented at the meeting of the Jean Piaget Society, Philadelphia, PA. Renshaw, P., &Garton, A. F. (in press). Children’s collaboration and conflict in dyadic problem solving. In C. Pratt, A. F. Carton, W. E. Thnmer, & A. Nesdale (Eds.),Research issues in child development. London: Allen & Unwin. Riegel, K. S. (1979). Foundations of dialectical psychology. New York: Academic Press. Rogoff, B. (1982). Integrating context and cognitive development. In M. E. Lamb & A. L. Brown (Eds.), Advances in developmental psychology (Vol. 2, pp. 125-170). Hillsdale, NJ: Erlbaum. Rogoff, B., & Gardner, W. (1984). Adult guidance of cognitive development. In B. Rogoff & J. Lave (Eds.), Everyday cognition: Its development in social context (pp. 95-116). Cambridge, MA: Harvard University Press. Rogoff, B., Gauvain, M.,& Ellis, S. (1984). Development viewed in its cultural context. In M. Bornstein & M. E. Lamb (Eds.), Developmental psychology: An advanced textbook (pp. 533-571). Hillsdale, NJ: Erlbaum. Rogoff, B., & Lave, J. (Eds.). (1984). Everyday cognition: Its development in socialcontext. Cambridge, MA: Harvard University Press. Rogoff, B., Malkin, C., & Gilbride, K. (1984). Interactions with babies as guidance in development. In B. Rogoff & J. Wertsch (Eds.), Children’s learning in the zone ojproximal development (pp. 31-43). San Francisco, CA: Jossey-Bass. Rogoff, B., Mistry, J., Radzisweska, B., & Germond, J. (in press). Infants’ instrumental social interaction with adults. In S . Feinman (Ed.), Social referencing and thesocial construction of reality. New York: Plenum. Rogoff, B., & Wertsch, J. V. (Eds.). (1984). Children’s learning in the zone of proximal development. San Francisco, CA: Jossey-Bass. Rosengren, K. S., Behrend, D. A., & Perlmutter, M. (1985, April). Pareni andchildspeech during puzzle solving. Paper presented at the meeting of the Society for Research in Child Development, Toronto, Canada. Rosenthal, T. L., & Zimmerman, B. J. (1972). Modeling by exemplification and interaction in training conservation. Developmental Psychology, 6 , 392-401.

Social Influences on Cognition

I43

Rozin, P. (1976). The evolution of intelligence and access to the cognitive unconscious. Progress in Psychobiology and Physiological Psychology, 6 , 245-279. Saxe, G. B., Geahart, M., & Guberman, S. R. (1984). The social organization of early number development. In B. Rogoff & J. V. Wertsch (Eds.), Children’s learning in the zone of proximal developmenf (pp. 19-30). San Francisco, CA: Jossey-Bass. Selman, R. L., & Yeates, K. 0. (1987). Social regulation of intimacy and autonomy: A developmental constructivist perspective. In W. M. Kurtines & J. L. Gewritz (Eds.), Moral development through social interaction (pp. 43-101). New York: Wiley. Silverman, I., & Geiringer, E. (1973). Dyadic interaction and conservation instruction: A test of Piaget’s equilibrium model. Child Development, 44, 815-820. Skon, L., Johnson, D. W., & Johnson, R. T. (1981). Cooperative peer interaction versus individual efforts: Effect on the acquisition of cognitive reasoning strategies. Journal of Educational Psychology, 13, 83-92. Sroufe, L. A., & Flesson, J. (1986). Attachment and the construction of relationships. In W. W. Hartup & 2. Rubin (Eds.), Relafionships and development (pp. 51-72). Hillsdale, NJ: Erlbaum. Stern, D. N. (1974). Mother and infant at play: The dyadic interaction involving facial, vocal, and gaze behaviors. In M. Lewis & L. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 187-214). New York: Wiley. Stevenson, H. W. (1982). Influences of schooling on cognitive development. In H. W. Stevenson & D. A. Wagner (Eds.), Cultural perspectives on child development (pp. 208-224). San Francisco, CA: Freeman. Strayer, F. F., & Trudel, M. (1984). Developmental changes in the nature and function of social dominance among young children. Ethology and Sociobiology, 5 279-295. Sullivan, H. S. (1953). The interpersonal theory of psychiafry. New York: Norton. Super, C. (1980). Anfhmpologicalperspectiveson child development. San Francisco, CA: Jossey-Bass. Syc, S. (1986). Problem solving alone and in pairs on a novel task: The Japanese abacus. Unpublished manuscript, University of Chicago, Chicago, IL. Sylva, K., Bruner, J. S., & Genova, P. (1976). The role of play in the problem solving of children 3-5 years old. In J. S. Bruner, A. Jolly, & K. Sylva (Eds.), Play: Its role in development and evolufion (pp. 244-257). London: Penguin Press. Tinsley, V. S., & Waters, H. S. (1982). The development of verbal control over motor behavior: A replication and extension of Luria’s findings. Child Developmenf, 53, 746-753. Turiel, E. (1966). An experimental test of the sequentiality of developmental stages in the child’s moral judgements. Journal of Personality and Social Psychology, 33, 611-618. Turiel, E. (1983). Domains and categories in social-cognitivedevelopment. In W. F. Overton (Ed.), The relationship between social and cognitive developmenf (pp. 53-89). Hillsdale, N J: Erlbaum. Vygotsky, L. S. (1929). The problem of the cultural development of the child. Journal of Genetic Psychology, 36, 415-434. Vygotsky, L. S. (1962). Thought and language. Cambridge, MA: Harvard University Press. Vygotsky, L. S. (1978). Mind and society. Cambridge, MA: Harvard University Press. Wagner, D. A., & Stevenson, H. W. (1982).Culturalperspecfives on child development. San Francisco, CA: Freeman. Wagner, M. E., Schubert, H. J. P., & Schubert, D. S. P. (1985). Effects of sibling spacing on intelligence, interfamilial relations, psychosocial characteristics, and mental and physical health. Advances in Child Development and Behavior, 19, 150-206. Webb, N. (1980). An analysis of group interaction and math errors in heterogeneous ability groups. British Journal of Educational Psychology, 50, 1-1 1. Wellman, H. M. (1983). Metamemory revisited. In M. T. H. Chi (Ed.), Trends in memory develop-

144

Margarita Azmitia and Marion Perlmutter

ment research @p. 31-51). Basel: Karger. Wertsch, J. V. (1979). A clarification and application of Vygotsky’s theory. Human Development, 22, 1-22. Wertsch, J. V. (Ed.). (1985). Culture, communication, and cognition: Vygotskian perspectives. London & New York: Cambridge University Press. Wertsch, J. V., McNamee, G. D., McLane, J., & Budwig, N. A. (1980). The adult-child dyad as a problem-solving system. Child Development, 51, 1215-1221. Wertsch, J. V., Minick. N., & Arm, E (1985). The creation of context in joint problem solving. In B. Rogoff & J. Lave (Eds.), Everyday cognition: Its development in social context (pp. 151-171). Cambridge, MA: Harvard University Press. Wood, D. J. (1980). Teaching the young child: Some relationships between social interaction, language, and thought. In D.Olson (Ed.), Thesocial foundations of language and thought (pp. 280-298). New York: Norton. Wood, D. J., Bruner, J. S., & Ross, G. (1978). The role of tutoring in problem solving. Journal of Child Psychology and Psychiatry, 17, 89-100. Wood, D. J., & Middleton, D. J. (1975). A study of assisted problem solving. British Journal of Psychology, 2, 181-191. Wood, D.J., Wood, H. A., & Middleton, D. J. (1978). An evaluation of four face-to-face teaching strategies. International Journal of Behavioral Development, 2, 131-147. Winegar, L. T. (1986, May). Role of adults in children’s understanding of social events. Paper presented at the meeting of the Jean Piaget Society, Philadelphia, PA. Youniss, J. (1980). mmnts andpeers in social development: A Sullivan-Piagetperspective. Chicago, IL: University of Chicago Press. Zajonc, R. B., & Marcus, C. B. (1975). Birth order and intellectual development. Psychological Review, 82, 74-88. Zaporozhets, A. V. (1969). Some of the psychological problems in sensory training in early childhood and the preschool period. In M. Cole & 1. Maltzman (Eds.), Handbook of contemporary Soviet psychology (pp. 86-100). New York: Basic Books. Zimmerman, B. J., & Blom, D. E. (1983). Toward an empirical test of the role of cognitive conflict in learning. Developmental Review, 3, 18-38.

UNDERSTANDING MAPS AS SYMBOLS: THE DEVELOPMENT OF MAP CONCEPTS IN CHILDREN

Lynn S. Liben DEPARTMENT OF PSYCHOLOGY THE PENNSYLVANIA STATE UNIVERSITY UNIVERSITY PARK, PENNSYLVANIA 16802

Roger M. Downs DEPARTMENT OF GEOGRAPHY THE PENNSYLVANIA STATE UNIVERSITY UNIVERSITY PARK, PENNSYLVANIA 16802

I. WHY MAPS? 11. OVERVIEW

A. THE PERVASIVENESS AND POWER OF MAPS B. DEVELOPMENTAL INQUIRY 111. WHERE HAVE WE BEEN? REVIEW OF PAST MAPPING RESEARCH A. DISCIPLINARY TRADITIONS B. CHILDREN AND MAPS: THE CONVENTIONAL WISDOM C. CONCLUSIONS 1V. THE MAPPING PROJECT AT PENN STATE (MAPPS) A. T H E STRUCTURE OF MAPPS B. T H E CHILD’S CONCEPT OF A MAP C. UNDERSTANDING CORRESPONDENCES BETWEEN MAP AND PLACE V. MAP UNDERSTANDING IN YOUNG CHILDREN REVISITED V1. MAPS AS SYMBOLIC REPRESENTATIONS VII. SUMMARY REFERENCES

145 ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR, VOL. 22

Copyright 0 1989 by Academic Press, Inc. All rights of reproduction in any form reserved.

146

Lynn S. Liben and Roger M. Downs

I F YOU’REWINGX, VISIT YOUR MOTHER IN NEEDLES, VOU’LL NEEDA MAP

AND NUMBERS AND NAMEMEAN

Reprinted by permission of UFS, Inc.

Miss Fulkes rotated the terrestrial globe until the crimson triangle of lndia was opposite their eyes. ‘That’s where Daddy and Mummy took the ship. Bombay is a big town in lndia,’ she went on instructively. ‘All this is India.’ ‘Why is lndia red?’ asked little Phil. ‘1 told you before. Try to remember.’ ‘Because it’s English?’ Phil remembered, of course; but the explanation had seemed inadequate. He had hoped for a better one this time. ‘There, you see, you can remember if you try,’ said Miss Fulkes, scoring a small triumph. ‘But why should English things be red?’ ‘Because red is England’s colour. Look, here’s little England.’ She spun the globe. ‘Red too.’ ‘We live in England, don’t we?’ Phil looked out of the window. The lawn with its Weliingtonia, the clotpolled elms looked back at him. ‘Yes, we live just about hers’ and Miss Fulkes poked the red island in the stomach. ‘But it’s green where we live,’ said Phil. ‘Not red.’ Miss Fulkes tried to explain, a s she had done so many times before, just precisely what a map was. (Huxley, 1928, p. 252)

I.

Why Maps?

Even the insights of Charles Schultz and Aldous Huxley might not prevent readers from skipping an article devoted to the development of children’s abilities to understand and use maps. The domain might appear isolated and esoteric and thus of little significance to general issues in cognitive development. In anticipation of this reaction, we devote the next section of the article to the question, “Why are maps important, and why should they be studied developmentally?’’ We believe that an answer to this two-part question is

Development of Map Concepts in Children

147

possible only by approaching it from an interdisciplinary perspective that integrates theory from cartography with theory from developmental psychology. Through the former, one can appreciate that maps can take a wide range of forms and serve a wide range of functions; through the latter, one can identify the cognitive skills that permit maps to be understood and exploited. Because most work to date has been conducted within single disciplines, Section 111 of the article is devoted to selective reviews of the work on children’s understanding of maps found within psychology and geography. We discuss the extent to which limitations of current disciplinary approaches to children’s map understanding may be generating a conventional wisdom that presents a simplistic and misleading picture of map understanding. In Section IV, we illustrate the alternative interdisciplinary approach by discussing our program of research on children’s map comprehension and production. We close the article (Sections V, VI, and VII) by setting both the mapping literature and our own work within a larger context by discussing ways in which theory and research on the development of map understanding interface with theory and research on the development of other modes of symbolic representation.

11. Overview A. THE PERVASIVENESS AND POWER OF MAPS

Maps are the preeminent graphic symbol system for representing the experience of space, pervasive across eras, cultures, and contexts. Historically, maps date from as early as 6000 B.C. at Catal Huyuk in Turkey (Harvey, 1980; Smith, 1987). Cross-culturally, maps appear in many forms, including wood carvings in Eskimo cultures, stick charts in the Marshall Islands, and designs on coins in ancient Greece. Within our own culture, maps appear in many forms and many contexts. They are common in wayfinding (e.g., road maps, shopping malls, airline route schedules, fire exit routes in hotels), they appear in print and television news reports, board games, children’s books, art, and legal reports. Although not necessarily well-known, the historical, cross-cultural, and contextual pervasiveness of maps is readily demonstrable to noncartographers. In contrast, the power of maps is more difficult to convey. Unfortunately, many people have restricted visions of maps, equating them with the traditional oil company road map, “the Gospel according to Rand McNally” (Wohlwill, 1973, p. 167). The restrictiveness of this view becomes readily apparent when maps are considered from the perspective of cartographic theory. Cartography reveals the vast range of forms of maps, including plastic relief maps that provide a three-dimensional topographic surface; orthophotomaps, or

148

Lynn

S. Liben and Roger M.Downs

map-photograph hybrids; satellite photographs with cartographic overlays; computer-generated maps; and perspective maps such as axonometric or isometric views of cities. (For examples of contemporary map forms, see Bertin, 1983; Monmonier & Schnell, 1988; Southworth & Southworth, 1982; Treib, 1980.) Modern cartographic technology permits a vast range of realizations of the appearance of portions of the earth’s surface, and of the distribution of phenomena (e.g., goods, services, or people) over that surface (Monmonier, 1985). In addition to being exposed to restricted forms of maps, children-and, indeed, many adults-have a restricted vision of the functions of maps. In social studies courses, maps are often presented as only archives of information, as factual storehouses that contain the right answer to questions such as, “Where is such-and-such a place located?” “How big is this place in relation to. . .?” In short, maps act as a simple graphic repository for information that might just as well be represented-albeit less parsimoniously and efficiently-in a propositional format, such as a gazetteer listing of city locations, or a table of the areas of political units. Although archival functions are important, they do not exhaust the uses of maps. Maps are creative statements about the world, not merely degraded reflections of it (Downs, 1981). The essential function of a map is rendering the experience of space comprehensible. As Treib (1980) argued, “maps are the projections of experience” (p. 20). The experience to be rendered may be direct and personal-the child‘s sketch map of the neighborhood-or indirect and impersonal-a satellite weather map-photo used in an earth science class. It may be of a space visible all at once-a map of a tabletop model-or of a space explored sequentially-a map of the school neighborhood. Whatever its specific form, the map is convincing because it captures the essential hypothetical property: “This is what the world would look like if. . .” In other words, the map brings the unperceivable extent of the world at large into perceivable bounds. The extended world is given a reality and comprehensibility that is otherwise difficult to achieve. The map, therefore, is a means of realizing an understanding of the world. Two senses of realize are implied. The first is realize in the sense of “making real,” of giving tangible form to something that was previously intangible. The map is a concrete thing that permits a series of manipulations: detailed inspection, rotation, extrapolation, interpolation, and so on. This sense shades into the second sense-that is, realize as in “understand.” One can comprehend relationships that were previously unsuspected but that become apparent through the structural properties of the map surface. Understanding follows from the appreciation of spatial relationship, continuity, coherence, scale change, perspective change, and so forth. We can make our fragmentary experience of the world comprehensible. These realizations do not depend

Development of Map Concepts in Children

149

on a literal, pictorial verisimilitude. The map is neither mirror nor miniature. It is a model of the world, a carefully controlled symbolic abstraction. Thus, although the map is an archive for storing large quantities of information efficiently, and although it can be used for wayfinding, the map offers far more. Maps suggest relationships not otherwise apparent, as when archaeologists use maps to link fragments of information and infer settlement patterns and paths of cultural diffusion, or as when epidemiologists use maps to infer mechanisms for the spread of disease. Maps are powerful means of persuasion, as when advertisers and propaganda agencies use maps to spur action. In short, maps are powerful means for making the world comprehensible, for posing and answering questions, and for persuading others to see the world in a particular way. Just as maps do not provide singular miniaturizations of the world, they do not provide undistorted renditions of that world. The representation of the three-dimensional earth by a two-dimensional map can preserve some, but not all of the four essential geometric properties of the earth’s surface: shape, area, distance, and direction. Thus, although maps can preserve the shapes of small parts of the earth’s surface, they cannot preserve the shape of a very large country. If area is of importance for a particular purpose, a map can be made that is “in scale” over its whole surface, but in this case, shape is distorted. One cannot have equivalence of areas and similarity of shapes at the same time. In short, contrary to popular opinion, no one “correct” map exists. Map forms differ and indeed should differ, depending on their function. Some maps, of course, serve navigation. But they may also be used for deriving or presenting measurements, not only for terrain information (e.g., elevation), but also for other data (e.g., population density, income per capita). Maps are used for visualization (e.g., as when one “sees” patterns in the spatial distribution of rainfall). Even more abstractly, the concept of a map is widely used as a metaphor for the internal representation of knowledge (Downs, 1981). B. DEVELOPMENTAL INQUIRY

If one accepts the argument that maps are a pervasive and powerful graphic symbol system, then the first justification for studying the origins and development of this particular system rests on the same grounds as the justification for studying children’s mastery of any important symbol system. As an example of symbol systems in general, maps raise questions about the nature of symbolic representation and the role of analogy. Indeed, the developmental study of mapping affords a particularly valuable domain in which to study the spontaneous development of symbolic functioning because unlike domains usually studied-language and mathematics-mapping is rarely instructed

I50

Lynn S. Liben and Roger M. Downs

intentionally at home or school, at least in the early years. Moreover, because of the particulars of this symbol system, maps also present an important domain in which to explore questions about the child’s ability to understand space, scale, geographic hierarchy, frames of reference, and interpolation and extrapolation. Apart from its utility for investigating symbolic and conceptual development, research on children’s developing understanding of maps is of interest from the perspective of education. This interest is especially obvious now, given that children’s and adults’ abilities to understand maps have been the focus of press coverage documenting a shocking level of “geographic ignorance.” Surveys, for example, have shown that “one in five 12-year-olds in North Dallas misidentified Brazil as the United States on a map of the world” (US. News & World Report, March 25, 1985, p. 50), and 25% of a college class in Wisconsin could not identify the U.S.S.R. on a map (Centre Daily Times, May 22, 1987). In order to develop sound geographic educational programs to remedy this ignorance, we must understand the origins, bases, and developmental changes in children’s understanding of maps. Importantly, if geographic education is planned within the context of a narrow view of map forms and functions, it will become little more than the endless memorization of locations (place learning). If, however, geographic education is developed within the context of a deep (and therefore flexible) understanding of the range of map forms and functions and of children’s potential for understanding these domains, it can be successful not only in teaching place locations and interrelationships, but also in enhancing understanding of symbolic systems more generally, spatial concepts, mathematical concepts, and other cognitive skills. Thus, the study of children’s developing understanding of maps can contribute to the enhancement of geographic education. Although this link is not the focus of the present article, it is discussed extensively elsewhere (see Downs, Liben, & Daggs, 1988).

111. Where Have We Been? Review of Past Mapping Research A.

DISCIPLINARY TRADITIONS

In this section, we review selected prototypical studies with children and maps that have appeared in the psychological and in the geographical and environmental journals, Although a selective review provides only a partial picture of past research, it does convey the distinct flavor of each disciplinary tradition, and presents specific examples for discussion. After illustrating these

De\>elopment oJ Map Concepts in Children

151

traditions, we describe the conventional wisdom that has evolved from them, and we offer theoretical and empirical critiques of that conventional wisdom. By highlighting strengths and weaknesses of past work, we argue that although research rooted within individual disciplines has much to offer, the contributions are limited and even misleading when viewed in isolation from one another. This argument provides the foundation for our interdisciplinary work on maps, which is discussed in the following sections. 1. Psychological

Work

a. Overview. In most research with maps in the psychological literature, maps serve as a dependent measure for externalizing some other phenomenon of interest. Using a taxonomy developed earlier (Liben, 1981), maps (a type of “spatial product”) are used as indicators of “spatial thought” or “spatial storage,” not as a focus of research in their own right. This generalization is especially true in adult cognitive psychology. Kosslyn, Ball, and Reiser (1978), for example, used adults’ performance on mapping tasks to investigate mental imagery. Levine, Jankovic, and Palij (1982) and Shepard and Hurwitz (1984) used maps to explore the role of orientation frameworks in spatial behavior. Hirtle and Mascolo (1986) presented information in map form to assess the impact of semantic labels on memory for spatial locations. McNamara, Ratcliff, and McKoon (1984) used the acquisition of knowledge from maps to explore the structure of the mental representation of knowledge. Their conclusion captures the role of maps in this research: “We want to emphasize that although priming [using maps] seems to be informative about certain aspects of spatial knowledge, it is only one of many tasks that might be used to study spatial cognition” (p. 730). This characterization of work in adult cognitive psychology likewise fits much of the work in developmental psychology. The majority of researchers using maps with children have been interested not in maps per se, but rather in some other construct that might be studied through the use of maps. For one subset of this work, maps are entirely incidental to the research endeavor. D. H. Feldman (1980), for example, used developmental progressions in children’s map drawings as a mechanism for studying a theory about stage progression and creativity. Maps were simply convenient-but not essential or even substantively relevant-for studying stage progressions. For another (and the largest) subset of this work, maps are more than mere conveniences because of the spatial nature of the questions addressed, but maps are still not the focal point. Piaget and Inhelder (1948/1956), for example, employed a range of tasks as a means of studying children’s developing spatial concepts. Mapping tasks provided one mechanism, but many others (e.g., haptic perception, perspective-taking, geometric sections, figure-copying tasks)

152

Lynn S. Liben and Roger M.Downs

were used as well. Hart (1979) used children’s performance on mapping tasks to reveal what children knew about their home and town environments; Siege1 and Schadler (1977) used models to investigate children’s knowledge of the spatial layout of their classrooms; Hazen, Lockman, and Pick (1978) used models to assess what children had learned about an experimental space; Liben and Newcombe (1988) used maps to assess whether barriers within a space affected children’s understanding of distance relations between points in that space; Gauvain and Rogoff (1986) asked children to produce maps to test whether different goals of exploration would affect children’s understanding of relationships within a space (for collections of other map-related work of this kind, see Cohen, 1985; Liben, Patterson, & Newcombe, 1981; Wellman, 1985).’ In all these instances, and in the bulk of research with maps and children, maps are useful as dependent variables because of their spatial nature, but they could be dispensed with if some alternative dependent measure were available. Although the bulk of the developmental research with maps and children may be characterized in this way, in a small subset of developmental research, maps are critical and could not be excised from the research endeavor without eviscerating it. Within this subset, three major approaches can be identified. The first approach is that of classic experimental child psychology. Investigators examine the success of children of various ages in performing some spatial behavior (typically going to a specified location within a room) as a function of the way that information is presented on a map (e.g., with the map flat on a table versus vertical; with the map inside versus outside the room; with maps with differing numbers of landmarks). Exemplars include work by Bluestein and Acredolo (1979); Presson (1982); Scholnick, Frank, Fein, and Schwartz (1986); and Uttal and Wellman (1987). The second is a case study approach, exemplified by the research conducted by Landau and her colleagues (Landau, 1986, 1988; Landau, Gleitman, & Spelke, 1981; Landau & Spelke, 1985). In an effort to identify the origins of spatial knowledge, Landau has explored spatial knowledge in a single child (“Kelli”), blind almost from birth. If Kelli could be shown to have spatial I In practice and in theory, distinguishing between models and maps is difficult. One might argue, for example, that models are three-dimensional and maps are two-dimensional representations. However, many maps are physically three-dimensional (e.g., maps molded from plastic that show elevation variations) and many models include two-dimensional cut-outs to indicate objects. Similarly, one might suggest that models and maps differ in scale, with models covering small areas of the earth’s surface and maps covering larger areas. Again, the distinction breaks down in practice: Many studies involve models of towns (a large area) and maps of classrooms (a small area). A fundamental, unaddressed question concerns the properties of different modes of graphic representation. Throughout this article, we concentrate on studies with maps but we include, where relevant, studies with models.

Development of Map Concepts in Children

153

knowledge, then visual experience must not be necessary for its emergence. Because another potential source of spatial knowledge derives from locomotion through a spatial environment and those inferential mechanisms that permit the calculation of spatial relationships, Landau studied Kelli’s ability to infer spatial relationships-not directly experienced-from locomotion. Because yet another possible source of spatial knowledge is through symbolic representation of space (e.g., maps), Landau studied Kelli’s ability to extract knowledge from maps. Thus, within Landau’s program of research, maps are one of several potential sources of spatial knowledge that are central to the question of the origins of spatial knowledge. The third is a profile approach illustrated by the work at Harvard Project Zero (Gardner & Wolf, 1987; Perry & Wolf, 1986; Wolf & Gardner, 1985). The driving aim is to catalog individual children’s developing symbolic systems (both verbal and graphic), and to understand the interrelationships among them. Of particular interest is how children become increasingly flexible with their symbol skills, becoming better able to select from a range of symbol systems, and to adapt symbols within a system, depending upon the purpose at hand. Among the systems studied-in addition to maps-are drawing, musical notation, mathematical notation, and verbal narratives. The focus of the Harvard Project Zero program is on symbol systems in general. Although each system brings unique characteristics of interest, any particular symbol system is optional. Maps are, therefore, as dispensable or indispensable as any other system studied. In summary, most work in which children have been asked to use or create maps has been driven not by an interest in maps per se, but rather by an interest in externalizing some other construct (e.g., environmental knowledge). A small but an important subset of developmental work has been concerned with maps qua maps. Three approaches have been identified: (1) the classic experimental child psychology approach; (2) the case study approach; and (3) the profile approach. Illustrative studies from each approach are described in turn below. In discussing these representative studies, we consider differences among them with respect to their goals, the kinds of methods used, and the theoretical orientations of the work.

6. Experimental Child Rsychology A widely cited study by Bluestein and Acredolo (1979) provides a good example of the work in the tradition of classic experimental child psychology. These investigators examined the ability of 3-, 4-, and 5-year-old children to use a map to find an object in a room. Of interest were the conditions under which the map was presented. Four conditions were formed by having the map presented inside versus outside the room, crossed with having the map aligned versus unaligned (rotated 180”) with the space. In a fifth condition, the map was held vertically outside the room.

154

Lynn S.Liben and Roger M. Downs

The session began with a familiarization phase, conducted inside a 12 x 12' collapsible room made with opaque curtain walls on a wooden frame and an oilcloth ceiling. The room contained a table in the center, surrounded by four chairs; distinctive features in each corner (an open-curtained door, a red box, a triangular-shaped foam object, and a curved-shaped foam object); and four boxes of different colors in the center of each wall. The child was seated at the table, on which was a map of the space in perfect alignment. The map, reproduced in Fig. 1, was described as follows: This map contained every distinctive feature of the room represented by colored line drawings from a vertical rather than aerial perspective, so that some pictorially represented three-dimensional information was provided to aid identification of the objects. The map also contained a separate representation of the toy elephant, with an adhesive backing, that could be repositioned on the map. (p. 693)

n

Fig. 1. The figure from Bluestein and Acredolo (19791, showing the experimental space.

Development of Map Concepts in Children

155

The child was asked to identify pictures on the map and point to their referents in the room, and was told that he or she was “to use the pictures to find Peanut the elephant” (p. 693). For map conditions inside the room, the child waited briefly outside the space while the experimenter hid the elephant and affixed the elephant sticker on the map; for map conditions outside the room, the experimenter left the child, entered the space, hid the elephant, and returned to place the sticker on the map. In each condition, the child sat at the table, examined the map, and went to retrieve the elephant. “In all cases the child was warned that the map might not be perfectly aligned with the room and to be very careful to choose the correct box” (p. 693). The success rates reported by Bluestein and Acredolo (1979) for all five conditions are shown in the top half of Table I. The Aligned-inside condition showed a significant increase in performance by age, with 100% of the oldest children succeeding on the task. Given the age difference in performance on the simplest condition, it is surprising to see that Bluestein and Acredolo reported no significant age differences for an apparently more demanding condition, the Aligned-outside condition. As they explained, however, the Aligned-inside condition served as a pretest and only those children who passed the pretest were tested under the four remaining conditions. Because the failures were not distributed evenly across ages (see the top half of Table I), the percentages reported by Bluestein and Acredolo do not provide normative

TABLE I Percentage Correct under Five Map Conditions Used by Bluestein and Acredolo (1979) for the Selected Sample (Original) and the Full Sample (Extrapolated) Condition Age

Alignedinside

Alignedoutside

3

55

75

4 5

86 100

92 100

3 4 5

55

41

86 100

19

Unalignedinside

Unalignedoutside

Verticaloutside

0 2s 90

67 7Ia 80

0 21 90

36 6fja 80

Selected Sample (Reported) 8 2s 80

Full Sample (Extrapolated)

100

4 21 80

“Given the original sample size of 24, the percentage reported should have been either 75 or 79%. Thus the extrapolated percentage should be either 64 or 68%.

156

Lynn S. Liben and Roger

M.Downs

data by age. Moreover, despite this selectivity by age, the differences in success by age on the Aligned-outside condition were still relatively large. We have extrapolated the percentages of children passing the various conditions using the reasonable assumption that children who failed the pretest would have also failed the other four conditions (see the bottom half of Table I). The conclusions drawn from the original and recalculated data are different. Bluestein and Acredolo argued that performance on the Alignedoutside and the Vertical-outside conditions was high and showed no age differences, whereas performance on the unaligned conditions (both inside and outside) was strikingly lower, with significant age differences, suggesting that only alignment is problematic for young children. The recalculated data, however, suggest that the problems of 3-year-olds and some 4-year-olds are not limited to lack of alignment between map and space. There were striking and apparently significant age differences emerging in the Aligned-outside and the Vertical-outside conditions as well. Apart from the particular conclusions drawn from these empirical data (a topic to which we return later), the study by Bluestein and Acredolo provides a clear example of the classic experimental approach. Studies in this tradition typically include a constrained set of task manipulations (e.g., the five map-presentation conditions), and use cleanly quantifiable measures of task performance (e.g., finding the hidden elephant). However, the theoretical orientation of this work as a whole is more difficult to characterize. A subset of this tradition, especially work associated with the Institute of Child Development at the University of Minnesota (e.g., Lockman & Pick, 1984; Pick, 1987), tends to involve greater emphasis on perceptual and motoric processes than on cognitive processes, perhaps stemming from a Gibsonian heritage. Even this generalization, however, is not a universal one. Acredolo (1981), for example, often links her questions to Piagetian theory. Although generalizing about the classic experimental child psychology approach is therefore difficult, the character of this approach comes into better focus when set against the two research approaches discussed next.

c. Spatial Knowledge: A Case Study Approach. The program of research by Landau and her colleagues provides a strong contrast to the experimental approach in its goals, methods, and theoretical orientation. At its broadest level, the work by Landau is aimed at revealing the origins of spatial knowledge. As part of that endeavor, Landau (1986) was interested in determining whether a very young child-in the absence of visual experience (the result of congenital blindness) and in the absence of prior experience with mapswould nevertheless be able to extract information from a tactile map to find the locations of objects in a room.

Development of Map Concepts in Children

157

To examine these skills, Landau (1986) asked Kelli, a 4-year-old child with whom she had worked on a variety of spatial and language tasks, to examine a tactile map. The map was presented under a variety of conditions in which the spatial relations varied among the child, the map, and the targets in space. In one condition, the map was directly in front of the child and aligned with the space. The targets were to the left, front, and right. In another condition, the map was placed to the child’s left or right, as were the targets, leading to either matched or unmatched map presentation and target position. A third condition was the front-behind analogue of the second condition, and a fourth condition was like the first condition except that the map was upright at approximately a 60” angle. The maps were gridded cardboard sheets and contained two wooden symbols, one representing the child and one the target (a toybasket or a toy). Kelli was tested in a 10 x 10’ room free of audible landmarks. She was carried into the room and placed in a chair. The description of the map was Kelli’s only prior map-reading experience: Kelli was shown the map of the room, and allowed to explore it. The map was described to her as she explored it, as follows: “This is a map of the room. It tells you where things are in the room. This is the whole room (guiding her hands over the entire map). And this is you, where you’re sitting in the room (guiding her hands to the block representing her). And this is the toybasket (guiding her hands to that block). Here’s where you are, and here’s where the toybasket is (touching each at the appropriate time).” She was asked to identify each symbol and did so easily. (p. 209)

Kelli was then asked, “Can you find the toybasket in the real room?” (p. 209). Sighted children were tested in an 8 x 10 x 4 ’ featureless enclosure that was contained within a room. The child was told that a toy would be hidden behind the walls of the “room,” was walked into the room with his or her eyes shut, and was seated in a chair. The description of the map followed the same pattern, with the exception of the haptic exploration. Children were asked, “Can you walk to where the toy is hidden?” (p. 210), and they then walked to the place in the surround behind which the toy was hidden. Performance was measured in four ways: the initial turn, the final position, fine directional accuracy, and gross directional accuracy. The proportions correct for each measure indicated that “the task was quite simple for the sighted children, with almost perfect performance by most subjects. Kelli also performed quite well, in all conditions” (p. 213). In contrast to the sighted children, Kelli’s performance did show slight improvement over trials. The results suggest that blind and sighted children can understand and use a simple map to guide navigation; that they can account for a left-right translation of the map in space; and that they can effect a front-back translation. Although both Kelli and the sighted children

158

Lynn S. Liben and Roger

M.Downs

were successful when the map was placed upright, “Kelli and two of the six sighted children initially interpreted the vertical presentation mode as representing a vertical space” (p. 218). They required feedback in order to counteract this belief. This case study offers some striking contrasts to the example of the classic experimental approach. First, Landau’s theoretical orientation favors biological capacities and predispositions. Concepts thought by Piaget to develop only gradually and constructively over years are seen as being present in infancy and undergoing little if any significant development. Landau and Spelke (1985) pose the possibility that there is very little development in the nature of the spatial knowledge system itself. Instead, developmental changes in spatial navigation seem to reflect an increasing coordination of the knowledge system with action, and with spatial markers in the world. That is, we suggest that development does not change the basic units of the spatial knowledge system; rather, children learn how actions and information in the world can reliably be used to locate particular objects in space. (p. 29)

Second, whereas those in the classic experimental approach tend to stay close to their empirical data, discussing which particular task manipulations have what particular effects at which ages, Landau has tended to go in the opposite direction, linking her particular empirical efforts to major issues that have intrigued philosophers and psychologists for centuries, such as relationships to Kantian, empiricist, and Gibsonian explanations of the origins of spatial concepts (e.g., Landau, Spelke, & Gleitman, 1984). Third, whereas the research procedures and analyses used in the classic experimental studies are typically well-established and uncontroversial, those of Landau are less so. For example, the measures used by Landau et al. (1984) to document Kelli’s ability to go “directly” to relevant landmarks have been criticized as permitting trial and error adjustments and allowing for a surprisingly large margin of error (Liben, 1988; Millar, 1988; but see Mandler, 1988, for a contrasting interpretation). Finally, whereas those researchers adopting the classic experimental approach typically base their conclusions on data from groups of children at particular ages, Landau and her colleagues base many of their conclusions on the intensive study of a single child, perhaps reflecting the child language tradition in which they are also involved (e.g., Landau & Gleitman, 1985). Even when comparison groups are studied (as in the aforementioned work in which Kelli’s performance was contrasted to that of 6 sighted children), the tendency is to be less concerned about keeping procedures precisely the same. Thus, unlike the procedure used in the Bluestein and Acredolo study in which the only variations across conditions were those of interest (Le., the conditions under which the map was presented), the blind versus sighted

Development of Map Concepts in Children

159

comparisons in Landau’s work involved differences not linked directly to visual experience. For example, Kelli was tested in a 10 x 10’ room, and the sighted children were tested in an 8 X 10 x 4’ featureless surround contained within a room. More importantly, test participation differed for the groups. “Kelli participated in four separate map experiments (henceforth ‘conditions’) conducted at different times. The sighted subjects each participated in a single (comparable) four-condition experiment” (p. 206). This difference might account for Kelli’s improved performance over trials and was apart from Kelli’s participation in many other spatial tasks conducted by Landau and her colleagues (e.g., see Landau & Spelke, 1985; Landau et a/., 1984). Furthermore, Kelli’s age ranged between 4 years, 9 months, and 5 years, 11 months, during the course of the study. For reasons of this kind, the comparisons between blind and sighted children offered by Landau are less compelling than the comparisons between conditions of map presentation or between children of different ages offered by those (e.g., Bluestein & Acredolo, 1979) working within the classic experimental child psychology tradition.

d. Harvard Project Zero: A profile Approach. Maps are one of a number of symbol systems explored in Harvard Project Zero to investigate children’s development and differentiation of graphic and verbal symbol systems. Although some general description of the map work in the context of Harvard Project Zero was provided by Wolf and Gardner (1985), the most detailed presentation was given by Perry and Wolf (1986). In the work on map production (Perry & Wolf, 1986), 39 middle- and lowermiddle-class kindergarten, first-, and second-grade children (approximately 5 , 6, and 7 years, respectively) viewed a model terrain. They were asked to make smaller 3-dimensional reconstructions of the terrain (which they did “with reasonable accuracy”), and then to make a map, showing “what each item in the town is, and where it is located” (p. 4). On the basis of these maps, Perry and Wolf (1986) concluded, encoding per se poses no problem. Asked to make a tree, a house or a road, all these children are able to produce very readable and even sophisticated renditions of the objects. But what develops more slowly is children’s ability to tune their record-making to the particular demands of map-making as compared to drawing. (PP. 4-5).

Specifically, Perry and Wolf suggested that the age-linked changes in and the ability to differentiate between maps and drawings show a number of trends in comprehensiveness, detail, symbol labeling, accuracy, orientation, proportion, and drawing angle. What characterizes both the Landau case study and the Project Zero profile approach is attention to the “big picture.” Unlike Landau’s focused approach, however, which relates empirical data from a constrained set of tasks in a

160

Lynn S. Liben and Roger

M.Downs

single (spatial) domain to broad theoretical issues of the origins of knowledge, the Project Zero group proposes to produce an integrated picture about diverse domains (here multiple symbol systems) by integrating empirical data from a wide variety of domains. Although this approach leads to far-reaching implications, no single domain is studied with the same detailed manipulation of factors as it is in the classic experimental approach. The Harvard Project Zero tradition typically involves testing relatively small numbers of children in multiple testing sessions with a variety of tasks. The resulting data illuminate the extent to which developing competencies in different domains are interrelated, and provide longitudinal data on development within and across domains. Landau also tested Kelli repeatedly over time, but she placed no emphasis on the consequences of Kelli’s changing age. Indeed, if anything, the fact that Kelli aged during the course of multiple studies seems to have been ignored.

e. Characterizing the Psychological Tfadition. Although the comments thus far have emphasized distinctions among the research approaches, these distinctions occur within a strongly shared psychological tradition. In all these research programs, procedures are described clearly enough to permit replication, coding is likewise valid and reliable, and statistical analyses are appropriately reported. The questions addressed are generally directed toward exploring individuals’ ability to use and/or create maps in relation to subject characteristics. Thus, not surprisingly, this psychological approach is characterized by close attention to variations in or manipulations of personrelated variables, and relatively little concern for variations in or manipulation of map-related variables. Indeed, the implicit and explicit picture of maps painted in most psychological research is narrow, unsophisticated, and potentially misleading. The map forms typically used are highly restricted. They are schematic, black-and-white line drawings. (The tactile map used by Landau [1986] is a pleasant exception, although this choice was presumably motivated by concern than by a concern for cartographic for a subject variable-blindness-rather variables-such as two- versus three-dimensional representations, forms of symbols, etc.) The maps used contain a minimum of information. In fact, a conscious effort was made to minimize map information. Presson (1982, p. 197) referred to Bluestein and Acredolo’s (1979) map as “rather detailed,” a surprising statement if one considers their map (see Fig. 1). Presson intentionally reduced the detail in his work, arguing that “the maps used in the current study had only one landmark that could be used to orient the map.” We are reminded of the Bellman in Lewis Carroll’s (Dodgson, 1939, p. 683) The Hunting of the Snark

Development of Map Concepts in Children

161

He had bought a large map representing the sea Without the least vestige of land: And the crew were much pleased when they found it to be A map they could all understand. “What’s the good of Mercator’s North Poles and Equators, Tropics, Zones, and Meridian Lines?” So the Bellman would cry: and the crew would reply “They are merely conventional signs!” “Other maps are such shapes, with their islands and capes! But we’ve got our brave Captain to thank” (So the crew would protest) “that he’s brought us the bestA perfect and absolute blank!”

The contents of the minimalist maps used in research are equally restricted, depicting small spaces (small rooms or, more commonly, subsections of rooms created with partitions). The rooms are “furnished” specifically for experimental purposes, in an impoverished manner when contrasted to real “living” spaces (e.g., a family room or a school classroom). Because these are room (or subroom) plans, they avoid some issues that are fundamental to mapping from a cartographer’s perspective-that is, issues involved in representing the three-dimensional surface of the earth on a two-dimensional surface in a geometrically systematic manner. The terrain used in the Harvard Project Zero work does include topographical features. However, because this terrain is a table-top model, it also avoids another classic cartographic issue in which a large space, incomprehensible in a single glance, is rendered comprehensible on a two-dimensional representation. As is logically implied by the observation that most researchers depict similar size areas with approximately the same size maps, map scale varies little in the psychological research. In addition to restricted forms of maps, spaces, and map scales, psychological researchers employ a restrictive range of map tasks. Subjects are almost invariably asked to use maps to find something (e.g., Presson, 1982; Scholnick et af., 1986) or, less typically, to archive information (as in producing a map of the model terrain in the Project Zero work). The restricted emphasis on the wayfinding function of maps is reflected in Bluestein and Acredolo’s (1979) definition of map reading as “the ability to make judgments about position in three-dimensional, large-scale space from information presented in a two- or three-dimensional small-scale representation of the space” (p. 691). But almost no attempts have been made to determine children’s abilities to interpret maps: to use maps to see patterns, to explore relationships, to generate or solve problems (Bruner, 1959, was an exception). Lobeck (1956), a cartographer, viewed map interpretation as

162

Lynn S. Liben and Roger M. Downs the process of reading between the lines of a story whereby the reader draws certain inferences and conclusions which the author did not specifically make. On a map, for example, the form of the hills, the location and pattern of rivers, the outline of the coast, and other details may provide the clue to certain geological information which the map makes no pretense of giving. (p. x)

Map interpretation involves a broader range of skills than simply the positional judgment that Bluestein and Acredolo (1979) specified. It includes, for example, interpolation and extrapolation, drawing inferences, and visualizing cross-sections. In addition to being restrictive, the psychological approach to maps may also rest on erroneous assumptions. An illustration of these assumptionsundoubtedly shared by other psychologists and by noncartographers more generally-may be found in the contrasts between drawings and maps described by Perry and Wolf (1986). Perry and Wolf began by noting that “In a drawing, a maker is free to select among landmarks and even to invent features. In a map, a maker must render all major landmarks and may not introduce new ones” (p. 7). This statement implies that maps show what is, in some fixed, singular manner. But maps do not reproduce the world (Downs, 1981, 1985). Cartographers select features to depict, a selection motivated by the purpose of the map. Thus, a cartographer might omit “major landmarks” of a city on a subway map and depict only subway stations. Furthermore, cartographers do “invent” features when they produce thematic maps that involve quantitative symbols (e.g., graduated circles) on a map of the United States to show distributions of cancer cases over a 3-year period. Cancer cases are not “there” in the same sense that a mountain or river is “there,” and yet thematic maps are maps nevertheless (Bertin, 1983). Similarly, a cartographer might take exception to Perry and Wolf‘s assertion that unlike a drawing in which the “maker can play with the spatial relations among objects and their orientations. . .in a map, the spatial arrangement of items and distances between them must not be distorted” (p. 8). First, insofar as one is mapping the three-dimensional earth’s surface onto a two-dimensional surface such as a piece of paper, distortion must occur. In producing a small-scale map, one decides which kind of distortion-area, distance, direction, or shape-is least serious for one’s purpose.2 But one of these properties must be distorted in the pursuit of preserving others. In this respect, distortions are necessary trade-offs. Various map purposes call for preserving different properties and hence suggest one kind of map projection 2The use of the term large scale here follows cartographic usage and refers to maps depicting a small area of the earth’s surface (e.g., a city map) whereas smallscale refers to maps depicting a large area (e.g., a world map). This terminology is often reversed in the psychological literature.

Development of M a p Concepts in Children

163

rather than another. For example, in sea navigation, angular relationships are critical (for setting the ship’s bearing). Thus, a Mercator projection might be used because it permits navigation along the same bearing. But a Mercator projection distorts area (hence the “Greenland effect,” which makes Greenland appear much larger than Brazil when it is, in reality, about the size of Mexico). If area is most important (as in showing farming acreage), a cylindrical equal area projection might be used. But a cylindrical projection distorts shape. (See Robinson, Sale, Morrison, & Muehrcke, 1984, especially Chapter 5, for a full and comprehensible discussion of projections and distortions.) Although such issues are not important for large-scale room plans, the expectation that “good” or “mature” maps do not distort is inaccurate. Furthermore, contrary to the suggestion of Perry and Wolf (1986), the portrayal of the “spatial arrangement of items and distances between them” may indeed be intentionally “distorted” or ignored on maps. Good examples are the popular and effective subway maps, which preserve information about relative order (which subway stop comes next), intersections (stations at which routes cross and transfers between lines are possible), and a sense of direction (which lines run roughly north-south and which roughly east-west), but, by a calculated omission, they make no attempt to preserve precise angular and distance relationships within and across subway lines. Also problematic is the contrast drawn by Perry and Wolf between drawings and maps: In a drawing, the maker has the option of playing with the way in which she represents the relative size of items or areas, in order to highlight the most interesting parts, or through the use of perspective techniques. This is not the case in maps, where proportional representation of sizes is crucial. (p. 8)

However, the cartographer does play with relative sizes of items and areas in order to highlight specific parts of a set of data. The symbols on a standard U.S. Geological Survey topographical sheet are not proportional. Road widths, as drawn, are clearly not in proportion to the actual scale of the map itself and often, roads of different capacities are not represented proportionally in series by width but by color. Map symbols for religious institutions, airports, and railway stations are not in proportion to their area on the earth’s surface. Moreover, in order to communicate particular relationships, mapmakers may play with the base unit of measurement. Thus, cartograms use proportional sizes of units (such as states or countries) based on population income, or any other factor of interest. The result is a map proportionally scaled to, say, income, rather than square kilometers. Furthermore, the use of perspective-drawing techniques is just as available to cartographers as to artists. Maps of urban areas frequently employ an affine geometry and a high oblique viewing angle to represent buildings and streets. Maps similar in

164

Lynn S. Liben and Roger M.Downs

construction to the now-famous New YOrkrS view of the world simulate a view from a high vantage point by decreasing area shown with increasing distance from the vantage point. Perry and Wolf (1986) also contrast drawings, in which “the point of view is typically from the front,” to maps, which “in the interest of preserving accuracy of spatial layout, most typically present a bird’s-eye view of both terrain and landmarks” (p. 9). Although it is true that the statement is qualified by “most typically,” their contrast suggests a view that orthogonal maps are somehow better, and that these are privileged in preserving accuracy. In short, Perry and Wolfs (1986) assumptions about sophisticated, mature mapping suggest fundamental misconceptions about maps, misconceptions that are probably shared by most psychologists, and indeed, by most noncartographers. Another problem with the psychological approach to maps is apparent in the presentation of maps used in research. One striking aspect of reports on children’s use of maps is the frequency with which figures showing the actual maps are absent altogether (e.g., Abel & Kulhavy, 1986; A. Feldman & Acredolo, 1979; Presson, 1982). Even when maps are discussed, critical map features are often ignored entirely or described vaguely, and the maps include puzzling features. Consider, for example, the description given by Bluestein and Acredolo (1979) cited earlier. It is not clear whether the figure of the experimental space provided in their paper (see Fig. 1) also depicts the map actually used with the children. They reported using “a vertical rather than aerial perspective” which suggests a sophisticated recognition of alternative mapping procedures. But they gave no discussion of such basic map elements as scale; why color was preserved; the choice of viewing distance; whether the viewing angle used in depicting the elephant was similar to that used for the furniture, or, more likely, was an elevation (side) view; the basis on which features are depicted in vertical rather than oblique perspective (e.g., boxes in Fig. 1 are shown in vertical perspective but foam shapes and walls are shown in oblique perspective); and more fundamentally, why it was necessary to use vertical perspective to provide “some pictorially represented three-dimensional information. . .to aid identification of the objects” (p. 693). The last point is puzzling insofar as Bluestein and Acredolo seemed to accept Blaut and Stea’s (1971) conclusion that young children readily understand the aerial perspective. Our discussion of problems in the conceptualization of maps has focused on room plans because, as we noted, they have dominated developmental psychologists’ research. Small-scale maps of larger areas have the potential for different kinds of errors and misunderstandings. Thus, if developmental psychologists extend their work to include maps of larger environments, as we believe they should, they might wish to turn to cartography as a source for map materials.

Development of Map Concepts in Children

165

2. Geographic and Environmental Work

a. Overview. Not surprisingly, the geographic and environmental work on maps has different paradigms and traditions, and relatedly, a different set of strengths and weaknesses. Cartographic researchers who study adults and children typically examine mechanisms for enhancing communication between the map reader and the map maker. Blades and Spencer (1986a) noted that two paradigms have dominated research on cartographic communication. The first paradigm, a psychophysical one, is concerned with the reader’s perceptual reactions to the map’s graphic properties. The classic study is that of Ekman, Lindman, and William-Olsson (1961) who derived a psychophysical constant, log n x 0.57, to express the magnitude scaling factor necessary to ensure that proportional circle symbols appeared to increase in area in the correct magnitude relative to each other. In the last few years, this psychophysical paradigm has been replaced by a cognitive paradigm that is focused on the processing of information. Eastman (1985), for example, explored the relationship between map learning and map form. He demonstrated that the graphic organization of the map has a strong impact on the regionalized groups of features (chunks) that are stored and that these chunks are in turn organized into a hierarchical memory structure. The work with children in geography is characterized by far greater attention to materials and tasks than that found in the psychological literature. Illustrative research programs associated with Clark University and the University of Sheffield are reviewed to provide specific examples. b. The Place Brception Project at Clark University. The Clark tradition is the work on children and maps that is undoubtedly best known to both geographers and psychologists. The studies, which began in the middle 1960s, were part of the Place Perception Project, aimed at examining environmental learning in young children. Although some of the concepts and personnel are rooted in psychology, we have classified this work as geographic, because the approaches and dissemination outlets are consonant with that categorization. The Clark investigations were directed toward (1) determining children’s abilities to understand aerial photographs and maps, to produce maps, and to use maps for navigational problems; (2) investigating the origins of these skills; and (3) developing classroom activities designed to enhance these skills. Unfortunately, although pieces of the research program have been described in accessible publications (Blaut, McCleary, & Blaut, 1970; Blaut & Stea, 1971; Stea & Blaut, 1973), much appears in relatively informal, nonarchival, research reports (e.g., Blaut, 1969; Stea, c. 1972; Stea & Blaut, c. 1970). Although some

166

Lynn S. Liben and Roger M. Downs

reports in the latter group contain more detail, the illustrative study is taken from the first group to ensure that the original source is available to the reader. Blaut et al. (1970) were concerned with whether children entering school “can interpret and utilize an environmental map without training or prior exposure to the representation’’ (p. 336). Arguing (1) that very young children cannot be tested on map-making tasks because they lack literacy and drafting skills, and (2) that map reading requires the same skills as map making, Blaut el al. (1970) used maps that make no literacy demands: “testing with a vertical aerial photograph-a map surrogate that has no legend (i.e., dictionary), no marginal information, and no other notation in the natural language” (p. 336). Regardless of whether one agrees that an aerial photograph is really a map, one must agree that some of the skills needed for the interpretation of aerial photographs are needed for map reading, and thus these data are relevant to understanding children’s mapping skills. For one sample (first-grade children, about 6 years old, from Massachusetts), the investigators began by showing the child a n oblique color photograph of a small town and countryside, “confirmed that the child was familiar with photographic images in general by acting out with the child the process of picture-taking,” and asked “What do you see in this picture?. . . Point to it” (p. 342). This question was repeated with a vertical, black-andwhite photograph of a residential portion of a town. The same procedure was followed with a second sample (first-grade Puerto Rican children), except that the oblique photograph was omitted. A subset of the children participated in a simulated navigation task. After roads and buildings had been identified on the vertical photograph, the child was asked to trace the outlines of a few roads and houses with pencil on acetate, and the photo was removed. After distracting the child for about a minute, the experimenter asked the child to name the traced figures, and then to color houses red and roads yellow. Next, a navigation problem was posed: Tivo houses were chosen and the child was asked to draw a pencil line showing the route he would follow to go from one of the houses to the other, by way of the roads depicted. The houses were selected so as not to be connected by a straight-line road; thus the task involved a simulated trip from one point to another one, which, in the real world, would not be visible from the first. A response was considered correct if it did not deviate seriously from the crudely drawn roads (p. 344).

The picture painted by Blaut el al. (1970) of the results is one that emphasizes young children’s competence. Virtually all (105 of 107) “perceived the vertical black-and-white photograph to be a downward view of a landscape and identified at least two unlike features in it” (p. 345). Performance on the navigation task was also good: 16 of the 19 children tested “performed the entire sequence of operations successfully” (p. 346). On the basis of their

Development of Map Concepts in Children

167

data, Blaut et al. (1970) concluded that preliterate children of school-entering age can (1) interpret a vertical aerial photograph, hence perform the mapping transformations of scale reduction and projective rotation, (2) abstract from the photographic image to a system of highly iconic map signs, (3) use the reduced, rotated, abstracted presentation in solving a simulated route-planning problem, and, therefore (4) engage in a real, if primitive, form of map reading, map making, and map use (p. 346). c. The ShefJield Research Program. A second relevant program of research that appears predominantly in geographic and environmental journals is that conducted in Sheffield, England, by Spencer (a psychologist) and his associates (e.g., Blades & Spencer, 1986b; Spencer & Darvizeh, 1981; Spencer, Harrison, & Darvizeh, 1980). In an article in Geography, for example, Blades and Spencer (1986b) reported on preschoolers’ abilities to use maps in a variety of ways. Of particular interest is their work on children’s abilities to use maps to find locations and routes in a room. Basing their work on Bluestein and Acredolo’s (1979) research discussed earlier (section III,A,l,b), Blades and Spencer (1986b) used a room “containing a selection of furniture’’ and six “hollow bricks” that served as hiding places for a small toy. At the door of the room, each of 16 3- to 4-year-old children was shown a 1:8 scale model of the room, furniture, and bricks (Leg0 blocks). The correct hiding place was pointed out on the model, and the child entered the room to find the toy. This procedure was repeated six times. In 83 of the 96 instances (16 children x 6 trials), children went straight to the correct brick. “The few mistakes which occurred were usually the consequence of the children’s enthusiasm to find the toy, which meant that they did not always look properly at the model before running into the room” (p. 48). From these data, combined, presumably, with data from other studies reviewed in their introduction, Blades and Spencer (1986b) concluded “that children’s success in these experiments suggests that young children are able to use maps and models of small environments to locate a particular place in that environment” (p. 48). In a variation of this procedure, with a room containing only a piece of furniture in each of the four corners, 20 children (3 to 4 years old) were asked to place a small doll in a model room to show the child’s own position in the room. “It was found that 16 children could understand the task, and could place the doll in the position which corresponded to their own position in the room” (Blades & Spencer, 1986b, p. 48). Tasks used to assess young children’s route-following skills were also included. Colored buckets were placed in the playground. In one layout, the buckets were positioned at different points; in a second layout, “paths (marked in chalk on the floor of the playground) were drawn between the buckets”

168

Lynn

S.Liben and Roger M.Downs

(p. 49). Maps for the first layout showed buckets by colored circles; those for the second also included the paths. Children were asked to follow a route indicated on the map, which they carried with them as they walked through the playground. Routes on the map either linked buckets directly (Layout 1) or followed the paths (Layout 2). As explained by Blades and Spencer (1986b): The experiment included a total of 17 maps showing different routes through the two layouts. Two groups of 7 children (mean ages 4 years 1 month, and 4 years 8 months) were asked t o use the maps. There was a very significant difference between the performance of these two age groups; the older group successfully followed the routes on the majority of the maps (mean number correct 13.7), but the younger group were unable to use many of the maps (mean number correct 4.1). In other words, the older children could be expected to follow the routes with few mistakes. The younger children seemed unable to understand the purpose of the maps-they sometimes walked straight to the target bucket, sometimes included all the buckets in a tour of the playground, or else they followed a route between the buckets which had no relationship to the route indicated on the map. (p. 50)

In a debatable step, Blades and Spencer (1986b) claimed that “since 4-yearolds had demonstrated the ability to follow simple route maps” (p. 50), additional studies were conducted with route maps for mazes rather than for rooms or playgrounds. Overall, then, Spencer and his colleagues also paint a picture of the competent preschooler. This emphasis on young children’s competencies is not the only quality shared with the Clark work. Discussed next are additional themes uniting much of the geographic and environmental work more generally.

d. Characterizing the Geographicand Environmental Tradition. The Clark work and the Sheffield work are typical of research concerning children’s mapping skills found in geographic and environmental journals. This tradition-unlike its counterpart in the psychological journals-typically includes diverse environments and diverse representations. Thus, within the two studies just described, materials included aerial photographs, room plans, scale models, playground layouts, and a simple cartographic map of a town with some symbolic elements (those produced by tracings of the aerial photograph). Similarly, the research involves a range of tasks. Although, as in the psychological work, some tasks were aimed at finding locations from a map, other tasks were aimed at indicating locations on a model; following routes from a map; showing routes on a map; and, importantly, assessing children’s understanding o r interpretation of what they were looking at (particularly in the identification tasks used by Blaut et al., 1970). Although these characteristics are commendable, this research tradition also has its flaws, which contrast to those of the psychological tradition. First,

Development of Map Concepts in Children

169

the descriptions of procedures are typically sketchy in the extreme. This lack of precision does not appear to be simply a problem in reporting, but instead suggests insufficient attention to, and perhaps control of, the procedures themselves. For example, Blaut et al. (1970) concluded that almost all children could “perceive the vertical black-and-white photograph to be a downward view of a landscape” (p. 345). However, as part of the introduction to the task, the tester “confirmed that the child was familiar with photographic images in general by acting out with the child the process of picture-taking” (p. 342). What precisely was acted out? Did it include a discussion or a pantomime of a downward view? Was the link between the photographic process and the aerial view made explicit? Evidence from work by Spencer et al. (1980), as well as from our own work (see Section IV,C), suggests that young children do not spontaneously understand that aerial photographs are photographs, or that they are taken from far away, or that they are downward views. The apparent inconsistency in findings may be attributed to the introductory material provided to the children in the Blaut et d.(1970) study, but the sketchy nature of the descriptions makes it difficult to evaluate this suggestion. In fairness to the geographic tradition, however, it should be noted that psychologists are not immune to the problem of providing children with information that is later assessed. Presson (1982), for example, was interested in “how children use map information [such as landmarks] to find a target when the map is rotated relative to the space” (p. 197). The procedure was described as follows: While the child watched, the ball was hidden in one of the containers and one map was marked while aligned with the space. The marked map was then rotated 90” and 180”, and the child was told he or she could still figure out where the ball was by noting where the landmark and the mark were in the map. The same procedure was followed with the other map before the test trials began. (p. 197)

These instructions explicitly suggest the use of a landmark strategy such that when children were later found to rely upon landmarks, the origins of that reliance are unclear. Thus, although the issue of confusion between instructions and dependent measures seems especially troublesome in the geographic literature, it is not entirely avoided in the psychological literature. A second problem that appears often in the geographic literature is the failure to specify procedures used to code data. In the Blades and Spencer (1986b) study, we are told that 16 to 20 children “could place the doll in the position [on the model] which corresponded to their own position in the room” (p. 48), but we are given no idea about the margin of error permitted. We are told that older children “successfully followed the routes on the majority of the maps” (p. 50), but not what behavior constituted success. We are told that in 83 of 96 instances, children went “straight to the correct hiding place” on the model-reading task, but we do not know what constitutes “straight.”

170

Lynn S. Liben and Roger M. Downs

Criteria for making judgments of this kind can be controversial. In their work with Kelli, Landau et al. (1981) described her travel to particular targets as direct, a characterization that has been challenged (see Liben, 1988; Millar, 1988). But the disagreement about how Kelli’s behavior should have been characterized is possible only because Landau el al. (1981), following accepted procedures in psychology, provided detailed information about scoring and data. Third, the data themselves are presented in little detail. Often only the means are given, with no sense of distributions, and with no indication of statistical significance. For example, the statement that “there was a very significant difference between the performance of these two age groups” (Blades & Spencer, 1986b, p. 50) was supported only by reporting the mean numbers correct for the two ages. Although the contrast in this instance (13.7 correct for children with mean age 4 years, 8 months, versus 4.1 correct for children 4 years, 1 month) might well prove significant, not all such contrasts are as clearcut. Fourth, in presenting conclusions, many investigators do not qualify their statements appropriately and thus often fail to provide a balanced picture. For example, despite the finding that the younger children (mean age, 4 years, 1 month) in the route-following task were correct on only 4.1 of 17 routes, two paragraphs after reporting these results, Blades and Spencer (1986b) stated, “Since 4-year-olds had demonstrated the ability to follow simple route maps.. .” (p. 50). They concluded that their experiments showed “that untrained children as young as 3 years old can use maps to locate places in small environments” (p. 52). Although even the young children were successful in finding the hidden toy, they did so using a model (not a map), in an extremely simple layout, and without need to consider alignment (which Bluestein & Acredolo, 1979, had demonstrated sharply impairs 3-year-olds’ performance). In light of these constraints, the conclusion by Blades and Spencer that 3-year-olds can use maps, although not technically incorrect, presents a misleading summary. Similarly, Blaut et al. (1970) concluded that children can solve a “simulated route-planning problem” based on success on a navigation problem. Although the child was asked to show how to go from one house to a second which, “in the real world, would not be visible from the first” (p. 344), the two houses were fully visible on the map itself, and the child was specifically told to go on the roads that had been outlined earlier. Thus, this task in essence required only that the child connect two points on a maze, and did not assess whether the child understood the representational nature of the map or its use for planning a route in a real space. In short, the problem with much work in the geographic tradition is that little thought seems to be given to what criteria (conceptual, measurement, or statistical) should be used to conclude that the child has actually mastered some skill or concept or indeed what skills are to be mastered.

Development of Map Concepts in Children

171

3. Psychological and Geographic Daditions Revisited The disciplinary traditions of psychology and geography have strengths and weaknesses in their contribution to our knowledge base about children’s mapping-strengths and weaknesses that become especially obvious when the traditions are contrasted. Investigations in the psychological literature are typically carefully designed, described, and analyzed; the conclusions usually illuminate the effects of controlled variations in constrained procedures, typically studied with respect to subject variables (often age). Despite these strengths, the studies are often restricted with respect to the questions addressed, particularly insofar as they involve highly restricted kinds of map forms and map tasks. Although ecologically valid in a narrowly defined sense, they lack a connection to the broader relationship between children’s mapping and the everyday environment. Even more importantly, many of the assumptions about maps that appear in the psychological literature are faulty. Investigations in the geographic and environmental literature have different strengths and weaknesses. On the positive side, they are typically addressed to a wider range of children’s competencies, in large measure as a result of an expanded range of stimulus materials and mapping tasks. Furthermore, incorrect assumptions about maps are rare. Unfortunately, however, this literature is simultaneously characterized by inadequate attention to specific procedures used for collecting, analyzing, and reporting data. Investigators have a penchant for presenting sweeping conclusions, not sufficiently qualified either with respect to unevenness in the data (e.g., concluding something about 4-year-olds in general when, in fact, the performance of younger and older 4-year-olds differed dramatically) or with respect to the limited nature of the data (e.g., concluding something about 3-year-olds’ abilities to use maps in general when the “map” is actually a model of an exceedingly simple space and is given only in an aligned condition). A conventional wisdom about the development of children’s mapping is emerging from these two traditions. In the next section, we characterize this conventional wisdom, and, in questioning it, we develop a basis for a discussion of our program of research. B. CHILDREN AND MAPS: THE CONVENTIONAL WISDOM

1. Characterizing the Conventional Wisdom

In reading both the psychological and the geographical literature, one might conclude that the development of map understanding in children occurs extremely early and extremely easily. On the basis of work with a 4-year-old congenitally blind child, Landau (1986) stated that “certain fundamental components of map use are accessible without specific prior experience in map

172

Lynn S. Liben and Roger M. Downs

reading, and without previous visual experience” (p. 201), and that primitive use of maps may occur by the end of the second year of life (p. 221). Similarly, on the basis of their work, Blaut and Stea (1971) concluded that “schoolentering six-year-olds can read and make maps” (p. 390), and that pilot tests with several three-year-olds. and with one perhaps precocious two-yearold, revealed that all were able to identify features on a vertical black and white photograph. If the photo interpretation ability emerges in certain children before they have learned to verbally describe their perceptions, then, since verbal report is precluded, unambiguous testing is very difficult and the adaptation of our original test design even more so. (p. 390)

In other words, map understanding appears so early that the date of its emergence is hard to pinpoint. Other researchers drew similar conclusions (Blades & Spencer, 1986b; Bluestein & Acredolo, 1979; Presson, 1982). On what basis do investigators account for the conclusion that children possess rudimentary mapping abilities so early and so easily? One form of the argument is that map use is simple because it is an automatic extension of the “spatial knowledge system” that is itselfsophisticated from a very early age. Landau (1986) took this position in arguing that: map use can arise independent of any specific experience in map reading. Why should this be so? One possibility is that core knowledge of maps is a direct and natural function of the spatial knowledge system required to support other kinds of spatial inferences seen in young children-specifically, those inferences observed in navigation tasks. (p. 220)

This beginning position thus shifts the question from identifying the origins of mapping to identifying the origins of the spatial knowledge system instead. Landau et al. (1984) suggested that this system arises easily because of the organism’s predispositions: We view the case of the blind child as an illustration of the problem any human faces in coming to know spatial properties of the world. While the blind child seems to be faced with a particularly impoverished environment from which to construct spatial knowledge, we believe that no environment is so rich as to impress a particular kind of spatial conception upon us, were we not prepared to embrace i t . . . . What is new is our evidence that spatial knowledge, like knowledge of language and knowledge of number, arises naturally in humans, with little thought or training and with no visual experience. (pp. 258-259)

In this argument, map comprehension is early and easy because it is a “direct and natural” (i.e., immediate and obvious) extension of “the early emerging spatial knowledge system” (Landau, 1986, pp. 219-220) that underpins the human ability to move in space. In turn, this spatial knowledge system results from a biological predisposition that prepares the human organism to use any environmental context (rich or impoverished) for development. An alternative answer to the question of why map skills are early and easy is that mapping is a simple extension of another ability that arises early and

Development of Map Concepts in Children

173

easily-perception. This argument rests on two premises: (1) understanding of maps is in large measure based on perceptual experience, and (2) even very young children have a wide assortment of perceptual experiences in their environments. Blaut et al. (1970) argued, In sum, the process of map-reading involves three transformations: one of perspective, rotation, or projection; one of distance, or scale; and one of semantic interpretation, from abstract sign to signification. . . . These three transformations are reminiscent of shape constancy, size constancy, and symbolic behavior, respectively, as these latter might appear at the level of macro-environmental events. Stated differently, the mapping transformations seem to involve an extension of the basic operations employed in learning the world of small objects. Indeed, if the relationship is close, then we might be led to predict that the two syntactic transformations, scale and projection, will emerge at an early age, since the constancies are developmentally primitive. (p. 339)

In short, the argument is that aerial photographs and cartographic maps reproduce the essential information available in the perceptual field-information that children have experienced repeatedly under a variety of conditions. This argument leads to the prediction that the young child, without instruction,

‘%anidentry and name anyfeature which the child would otherwise recognize from an earthboundperspective” (Blaut, 1969, p. 26; emphasis in the original). By implication, both of these answers require an assumption that maps are transparent. That is, in either answer, one is presumed to see through the surface of the map immediately to the world that it represents. The surface form of the map-that is, its graphic and geometrical properties as a representational medium-does not intervene in the process of map understanding (Downs & Liben, 1988). If maps are indeed transparent, then a reasonable supposition is that their interpretation should be early and easy, and thus, as Landau (1986) stated, “Appreciation of explicit representations of space seems to develop in children with little formal training” (p. 219). Taking the conventional wisdom to the extreme, one is left with little to account for or explain. From one perspective (Landau’s), the interesting questions concern the origins of spatial knowledge; from the other (Blaut & Stea’s), they concern the development of perceptual constancies. Once one overcomes the initial surprise in finding 2-year-olds reading maps, the question of the development of map understanding in children has been explained away.

2. Questioning the Conventional Wisdom The conventional wisdom might be questioned on two bases-one theoretical and the other empirical. At the theoretical level, one challenge stems from the constructive view of cognitive development offered by Piaget. The conventional wisdom that reading maps is an early and easy accomplishment implies that even very young children can appreciate the representational nature of maps, and can appreciate the spatial information contained in them.

174

Lynn S. Liben and Roger M. Downs

Yet Piagetian theory identifies important limitations in young children’s representational and spatial concepts. Although Piaget proposed that understanding the general notion of representation-that one thing can stand for another-marks the transition from sensorimotor to preoperational thought and thus may be expected to be in place by 2 years of age, he did not assert that representational deveiopment is complete at that point. On the contrary, representation during the preoperational period is characterized by being inflexible and concrete. According to this theoretical position, preschoolers have difficulty in maintaining the essential separation between representation and referent. Consistent with this position are findings that young children treat a representation as if it were the referent itself (e.g., as when young children lick a photograph of an ice cream cone, Beilin, 1983; or turn a picture over to see the back of a depicted object), or seem unable to ignore the qualities of the representation long enough to permit appreciation of its representational nature (as when the 2;year-old child has difficulty in understanding that a scale model of a room is simultaneously an object in its own right as well as a representation of the full size room, see DeLoache, 1987). Similarly, Piaget and Inhelder’s (1948/1956) position that children develop topological and then projective and Euclidean concepts gradually throughout childhood, suggests that the interpretation of the spatial properties of maps should be neither early nor easy. Projective and Euclidean concepts of space must be invoked to understand how to align maps and their referents; to appreciate distance and angular information; to understand the issues involved in projecting three-dimensional space onto a two-dimensional plane, and so on. Cartographic theory is also at odds with the early and easy conventional wisdom. It calls into question the conclusion that maps are transparent. Cartographic theory demonstrates the virtually limitless alternatives for representing the same place, the existence of a range of map functions, and the symbolic nature of the medium. The myriad of possible representations, the role of convention, the alternative projection systems, and other variations in map forms and functions, combine to demonstrate that maps cannot be transparent, singular miniaturizations of the world, and thus cannot be automatically and immediately interpreted by the visual perception system. At the empirical level, the criticisms already raised in discussing the psychological and geographical traditions imply that the conventional wisdom is incomplete and often misleading. As discussed earlier, the range of maps, tasks, and spaces has been limited. The age ranges of children tested in past research have been limited. Most research has been focused on very young children who are, indeed, appropriate for investigating children’s understanding of the representational nature of maps, and their ability to apply elementary spatial concepts to maps (e.g., the use of landmarks for orienting oneself on

Development of Map Concepts in Children

175

a map; the need to consider viewpoint in aligning self, map, and space). However, given that projective and Euclidean concepts undergo significant development during middle and late childhood, older children should also be studied to investigate more advanced aspects of mapping, including the use and implications of alternative projection systems when transforming three-dimensional spatial information to the two-dimensional plane, or the use of various coordinate grid systems for locating points or areas in space. Furthermore, with few exceptions (e.g., Perry & Wolf, 1986; Wood, 1977)’ the data are cross-sectional rather than longitudinal. If we are to understand the progressions and the interweaving of general cognitive progressions and the development of map understanding, then we must employ longitudinal research designs. The need to examine skills and changes within children is especially apparent when one considers the individual differences that abound. For example, Bluestein and Acredolo (1979) found that half of their 3-year-olds used a room plan map successfully in the simplest (Inside-aligned) condition. Half did not. Uttal and Wellman (1987) found that of the 4- and 5-year-olds who had been shown a map before traveling through an experimental space, only about a third knew more about locations in that space than did children not shown the map. Little attention has been given to the foundations and implications of individual differences of this kind. The general tendency is to dismiss them by incorporating them within the error term rather than by treating them as an interesting phenomenon to be studied in their own right (Downs, 1981), a tendency that may be dysfunctional for identifying the bases of successful mapping performance. Another fundamental difficulty with the empirical literature concerns the issue of appropriate indexes of map understanding. First, few investigators have explicitly distinguished between map reading and map interpretation. Second, the criteria used as evidence of map understanding may be too lenient. One conclusion of the Clark work was that very young children demonstrate remarkable competencies in understanding two-dimensional representations of the earth’s surface, a conclusion that rests on children’s success in picking out a few items correctly. These investigators offer no data on the occurence of identification errors, however. In a more balanced report of successes and errors on a similar aerial interpretation task, Spencer et al. (1980) reported that although the preschoolers were successful at recognizing some geographical features, especially those like roads that “retain sufficient of their ground-level characteristics to be recognized in an aerial view” (p. 61), the children’s interpretations were clearly limited. None of the children successfully recognized tennis courts in the aerial photograph; some even suggested that the tennis courts were doors. Apparently, children “found no inconsistency in seeing the lines as the panels on what would have been a

176

Lynn

S.Liben and Roger M.Downs

pair of huge doors lying flat beside the minute blobs that they had previously identified correctly as being full sized trees” (p. 62). We have found similar kinds of errors in our work (described in Section IV,C). When only successes are reported, young children do appear remarkably competent. When the successes are reported in the context of the errors, however, a different picture emerges. A systematic analysis of types of errors might lead to a more balanced understanding. Another serious problem is in the reporting of results within and across programs of research. First, investigators appear to modify the impression of their own findings as they retell them. Thus, Spencer and Darvizeh (1981) summarized earlier work (Spencer el al., 1980) as showing that “children as young as three years of age. . .found few problems in interpreting photographs comparable with those used by Blaut and Stea” (p. 23). Although one could quibble over whether the errors were few or many, the impression generated by the kinds of errors that did occur (e.g., tennis courts interpreted as doors) suggests important qualitative differences between children’s and adults’ interpretations, an impression that is lost in this summary statement. The problem of reporting across studies is even more striking. For example, despite the fact that in the Bluestein and Acredolo (1979) study, half of the 3-year-olds failed the simplest map-reading condition and were therefore dropped from the remainder of the conditions, this study has been cited as having shown that “Most three-year-olds were able to perform these rudimentary map reading exercises” (Spencer & Darvizeh, 1981, p. 24). C. CONCLUSIONS

Although each research tradition offers much of interest and holds many ideas relevant to children’s mapping, the conventional wisdom that is growing from past literature has some important limitations. We believe that coherent theories from both cartography and developmental psychology must be combined in designing research in this domain. In our program of research on children’s mapping, we have combined these perspectives, and for that reason, we describe a part of this work in the section that follows. We are well aware that some of the criticisms raised with respect to others’ work (e.g., restricted ages; lack of longitudinal data) apply equally to ours. We offer it, however, as an attempt to address some of the concerns arising from earlier work.

IV. The Mapping Project at Penn State (MAPPS) A. THE STRUCTURE OF MAPPS

In this section, we present a general description of our program of interdisciplinary research. We believe that the early and easy conventional wisdom

Development of Map Concepts in Children

177

provides an oversimplified view of the development of map understanding. We offer in its place a view that the development of map understanding is (1) gradual; (2) complex and multifaceted; (3) dependent on cognitive level and experiences and thus displays major individual differences; (4)coherent in that the trends in mastery of map understanding can be predicted and interpreted from a joint developmental and cartographic perspective; ( 5 ) replete with important educational implications; and (6) linked to the larger issue of the development of symbolic representation, a point expanded in the final section of this article. Given the importance of the interdisciplinary framework, we begin with a review of the basis of the framework, then outline our research program, and finally discuss illustrative findings to characterize what we believe is the essential nature of the development of map understanding in young children.

1. Interdisciplinary Foundations of MAPPS As explained in Section II,A, the cartographic component of our position is built on the idea that any map is a creative statement of belief about the world. The map renders the experience of space comprehensible in new ways. In realizing an understanding of the world, map makers and map readers must understand two basic correspondences that control the map-world relationship. Representational correspondences specify the information to be presented (abstraction and generalization) and the graphic means for its portrayal (symbolization), while geometrical correspondences specify scale (or viewing distance), viewing azimuth, and viewing angle relationships. As creative statements, maps are as flexible a form for the expression of ideas as any natural language. Just as language can become clichid, so too can maps. The form of maps is limited only by the technical skills and imaginative abilities of the map maker. Just as language serves many functions, so do maps. Our research begins by acknowledging the flexibility and breadth of maps as a representational system. Our work is simultaneously driven by a concern for the ways in which the understanding of maps rests on the cognitive skills of the map user, skills that we believe undergo significant developmental changes from early childhood through adolescence. First, at the heart of the ability to understand and to use maps is an understanding that one thing can stand for another; this understanding is achieved gradually during childhood. Maps are neither mirrors nor miniatures (Downs, 1985): they are analogical models of the world and therefore subject to the processes of analogical reasoning (Brown, 1989). Second, an understanding of maps requires an appreciation of logical relationships, and is thus dependent upon the developing child's logical thought processes. For example, a child's failure to understand class inclusion (that items in a particular subclass are simultaneously subsumed

178

Lynn S Liben and Roger M. Downs

by a superordinate class, see Piaget, 194111965, Chapter 7) may interfere with that child’s understanding of the geographical hierarchies implied on a map (see Downs et al., 1988). Third, an understanding of maps requires a comprehension of geometric relationships, and is thus dependent upon the child’s developing spatial concepts. As a representation of a space in space, the map possesses an intrinsic geometry. As a representation of some other space, the map shares an extrinsic geometry specifying its projective correspondence with the original space. Understanding space and geometry is, therefore, crucial to map understanding. Thus, a child who has not yet mastered the concepts of projective space (as assessed, for example, by a Piagetian perspective-taking task, see Piaget & Inhelder, 1948/1956, Chapter 8), would be expected to have difficulty on a mapping task that requires an understanding of the correspondences among alternative map views created by changing viewing angles and azimuths. Finally, individual differences exist in levels of performance on mapping tasks. Again from a Piagetian perspective, stage, not age, is more important in accounting for individual differences. Individual differences are also important in their own right because they reflect the impact of the available cognitive structures (especially those related to logical operations and concepts of space) on the child’s understanding of maps.

2. General Goals and Procedures of MAPPS The purpose of our project is to learn about the development of children’s understanding of the conceptual, functional, and formal properties of maps, and of children’s ability to use maps. Specifically, we are interested in map conceptualization (the understanding of the concept of a map, including its essential character and functions), map identiJication(the understanding of the formal components of a map, such as scale, projection, and the referents of symbols), and map utilization (the ability to use maps, as in wayfinding and pattern recognition). In addition to relating accomplishments in these areas to age as a means of providing normative data, we examined mapping skills in relation to cognitive variables, especially those drawn from Piagetian theory. This article highlights findings from two data sources. The first source is interviews with young children. The interviews were conducted with 3- to 6-year-olds enrolled in preschools. Each child was shown a sequence of place representations and was asked questions about map conceptualization, identification, and utilization. The selection of representations was designed to vary the medium (aerial photograph versus map); the familiarity of place (local versus distant); the viewing angle (vertical versus low oblique); the graphic representation system (e.g., perspective drawing versus conventional road map

Deselopment of Map Concepts in Children

179

format); the color (black and white versus color); and the graphic symbol system (highly abstract versus iconic). Included were an aerial photograph of downtown Chicago (vertical black-and-white photograph with an approximate scale of 1:12,000); a similar aerial photograph of the children’s local State College, Pennsylvania area (scale approximately 1:20,000); a Rand McNally road map of Pennsylvania; a tourist map of downtown Washington, D.C.; a perspective black-and-white view of central Philadelphia; and a matched aerial photograph-map set of downtown Chicago. The interview was designed to elicit children’s spontaneous comments, as well as their responses to a continuum of increasingly specific probes. Videotaped interviews were scored by two independent raters who transcribed and categorized responses. Discrepancies were resolved through joint discussion with a third judge. The second source of data was derived from testing conducted in an elementary school serving a heterogeneous community in a small town in a rural region of central Pennsylvania. We worked with 11 kindergarten through second-grade classes, enrolling a total of about 265 children ranging in age from 5 to 8 years. The classroom activities served a dual function: They provided an opportunity to collect data as well as to conduct an instructional program on mapping. Thus, on occasion, some procedures were not ideal for one purpose or the other (e.g., pedagogical requirements led to fixed rather than counterbalanced or randomized task orders; research requirements led to some limitations on class discussions). As was the case for the preschool interviews, the selection of materials was driven by cartographic theory. Maps varied in scale, familiarity of content and area, representational mode, and principal function. Mapping tasks were similarly varied. Children were asked to depict overall configurations, to mark specific locations and routes, and to translate directions from an actual space or a representation of a space to a map. Map-to-space relationships were systematically varied with respect to the immediate perceptual availability of the space to be mapped and with respect to the alignment relationships among self, map, and space. In addition, we collected data from individual children for whom we had obtained signed parental consent (about 75% of the sample). To explore the role of individual differences in map understanding, we gave Piagetian spatial tasks, spatial tasks taken from the psychometric tradition, and standard scholastic achievement measures. Given the scope of the project, a complete review of the findings cannot be presented here. Instead, we focus on a subset of the data to address two central questions: (1) From a child’s perspective, what is a map? and (2) How do children understand the essential correspondences (representational and geometrical) between a map and a place?

180

Lynn S. Liben and Roger M. Downs

B. THE CHILD’S CONCEPT OF A MAP

A map is a generalized, reduced, symbolic spatial representation of reality that has been transformed from the spherical surface of the Earth (or any celestial body) in some dimensionally systematic way. The preceding answer to the question, “What is a map?” is one given by a cartographer. But what answer might a child offer? How might that answer change with cognitive development and with increasing exposure to maps? In short, what is the child’s concept of a map? The key to the cartographic definition is the phrase “spatial representation of reality.” The stundfor property relates the map and its referent. When and how do children understand that a pattern of lines and colors or a pattern of gray tones stands for a particular place? As discussed earlier (see Section III,B,2), cognitive-developmental theory suggests that the stand-for relationship is not easily appreciated by young children. In particular, the Piagetian position is that representational thought is a gradual accomplishment. The transition from knowing things through action to knowing through representation marks the transition from sensorimotor to preoperational stages. Importantly, however, during the preoperational period, the logical thought processes that can be brought to bear on representations are significantly limited. One major thrust of our research is to explore the extent to which children interpret a place representation as standing for a place. We consider this issue at two levels. One, the holistic level, deals with the child’s understanding of the relation between the referent (space) and the symbolic representation taken as a whole (the map). The other, the componential level, concerns the child’s understanding of the relation between elements of the representation and elements of the referent (for example, road symbols and roads; forest symbols and forests). In the remainder of this subsection of the article, we discuss evidence concerning the holistic stand-for relationship; in the next subsection, we present evidence concerning the componential stand-for relationship. The responses of 3- to 6-year-olds to the Chicago aerial photograph and the Pennsylvania road map provide an informative contrast concerning the contents of the child’s map category. The Pennsylvania road map elicited confident, assertive responses, and children often said, “I know what this is; it’s a map!” before the experimenter could ask, “Could you tell me what this is?” In response to this direct question, children characterized the road map in terms of its content and its form (“That’s a map,” “States and stuff’). In contrast, the Chicago aerial photograph was almost always characterized only by its content (“Lots of buildings,” “City”), and only occasionally by its form (‘A picture”), although never a “photograph.” Although both representations appeared to be understood as showing places in general, neither

Development of Map Concepts in Children

181

elicited a uniformly reasonable understanding of particular places. The road map, for example, was said to show “Indiana, Pennsylvania,” “Part of Africa,” and “California, Canada, the West, and the North Coast”; the aerial photograph was said to show “the USA” and “the whole world.” When asked about the differences between the Chicago aerial photograph (presented first) and the Pennsylvania road map, children almost always answered on the basis of surface form, color in particular (“It has more colors”). Although some representations were readily understood as representations of a place by even very young children, others-about which adults would express no doubt-were not so interpreted. A tourist map of Washington, D.C., for example, confused many children, who called it a “cage” or a “space ship” or failed to interpret it at all. These observations suggest that the particular form of a place representation is indeed significant and that map understanding is not uniformly early and easy. To obtain data bearing more directly on what representations fit within children’s and adults’ category of map, we showed children and adults slides of a variety of place representations, and asked their opinion about whether each slide showed a map. Response categories were “Yes” (“I think this is a map”), “No” (“I think this is nor a map”) or “?” (“I’m unsure or I can’t decide whether this is or is not a map”). The representations were selected on the basis of cartographic theory. We varied such key map characteristics as medium of presentation (e.g., photographs versus drawings); place represented (e.g., local area versus the surface of another heavenly body); spatial scale (small scale through large scale); viewing angle (e.g., orthogonal versus oblique); color (color versus black-and-white); content (e.g., topographic versus thematic maps); technical graphic properties (e.g., hachures versus contour lines); and presence or absence of verbal labels. Combinations of characteristics lead to many types of place representations. We sampled 20 types and chose 3 exemplars of each type, yielding three 20-item sets of slides. Because children’s attention spans are limited, only one set was used with the youngest children, and thus the data discussed are drawn from these 20 slides only. The limitations of attention spans and nonindependence of some map characteristics prevented complete crossing of all factors, and therefore the conclusions offered are not definitive. Children and adults were almost unanimous in categorizing as maps those representations that showed places in small- to medium-scale, with color, from directly overhead, and with conventional cartographic symbols (Downs 8z Liben, 1988). These representations had much in common with the type of objective standard map that Wohlwill (1973, p. 167) captured in the phrase, “the Gospel according to Rand McNally.” Deviations from this profile of characteristics increased the likelihood that subjects, particularly young

I82

Lynn S. Liben and Roger M. Downs

children, either would say that the representation was not a map or would report that they were uncertain. Black-and-white and perspective views were more likely to be rejected. Large-scale place representations were likely to be rejected. Representations that were either photographic or pictorial were also likely to be rejected. Thus, an eye-level, color photograph of the Chicago skyline was decisively rejected as a map by all age groups. However, with increasing age, the map concept expanded to encompass a wider range of place representations. A black-and-white perspective view of central Philadelphia was increasingly accepted as a map, despite its departure from the profile of typical map characteristics in several ways (i.e., it was large-scale, partially pictorial, and not an overhead view). Some departures caused more disagreement than others. For example, the photographic nature of place representations such as a Landsat satellite image of Southeast England and an Apollo photograph of Earth resulted in mixed responses at all ages. In the case of a gray-tone, plastic-shading relief map of Italy, children and adults diverged in their assessments, the majority of adults accepting it and the majority of children either rejecting it or not being sure. Whether this change reflects increased acceptance of the particular map form (gray-tone, plastic-shaded, unlabeled), or increased recognition of the boot-shaped peninsula, is unknown at present. Although not definitive with respect to all questions, the data do demonstrate that the child’s ability to understand the holistic stand-for relationship develops slowly and in complex ways. It is not an accomplishment that emerges uniformly early and fullblown. Children can indeed understand the basic stand-for relationship-that the representation shows a placeperhaps even as early as 3 years of age. They can indeed distinguish among forms of place representations (e.g., maps versus pictures), and largely share a prototypical map concept with adults. Once established, the map concept changes with age and, presumably, experience, so that it increasingly encompasses a broader range of map forms. To what extent is this change a function of exposure to novel map forms? Does the graphic form determine incorporation into the map concept? Although we cannot answer these questions at present, we can state that maps are not immediately transparent, and that the representational medium can obscure the message (Downs & Liben, 1988). Our belief that the understanding of maps is not immediate is supported by data on children’s understanding of the correspondences between a map and a referent space. We now turn to these data. C. UNDERSTANDING CORRESPONDENCES BETWEEN MAP AND PLACE

At a componential level, two sets of correspondences determine the form of a map: (1) Representational correspondences produce the generalized,

Development of Map Concepts in Children

183

reduced, symbolic spatial representation, and ( 2 ) geometric correspondences create the reduced representation that has been transformed in some dimensionally systematic way. The process of understanding correspondences is shaped by the interactions among three factors. The first is the role of context: To what extent can the child appreciate the systematic interrelationships among the elements of a map and thus not interpret elements as discrete and isolated? The second is the role of iconicity: To what extent does the child recognize map elements as analogues that must be interpreted symbolically and not literally? The third is the role of convention: To what extent does the child recognize that the graphic form of specific map elements is arbitrary and thus neither inevitable nor inflexible? Understanding the interrelationships between context, iconicity, and convention is a lengthy and difficult achievement. In struggling with representational correspondences, the child must confront the stand-for relationship at the componential level. In our interviews, we found that although children could appreciate the stand-for relationship at the holistic level (as discussed in Section IV,B), they had difficulty in consistently maintaining the idea of a symbolic representation, and thus they interpreted map elements for what they looked like. Sometimes this strategy was advantageous. For example, in the Chicago aerial photograph, they recognized roads as roads “because they are gray”; in the Pennsylvania road map, they identified the Susquehanna River as water because “it is blue.” But more commonly, these “look like” interpretations worked against map understanding. Thus, in the Chicago aerial photograph, a baseball diamond was said to show “a guitar” and “an eye.” In the Pennsylvania road map, the Rand McNally compass rose symbol (located on Lake Erie), was said to be “the sun,” “a basketball stadium,” “feathers,” and, perhaps reasonably (because it is “in” water), “the place where the lifeguard sits.” Our work also reveals evidence of reification-that is, of an overextension of the symbol to the referent. In the interview study, children believed that a road shown in red on the map of Pennsylvania would actually be red in the real world; an airplane (symbolizing an airport) indicated a single airplane in the real world, and once it flew away, the symbol would no longer be used on the map. Reification was evident not only in the preschool interview data, but also in the classroom studies. Children were given a map of their classroom and were asked to mark locations of objects that were not already depicted on the map. In marking the location of a circular wall clock, children frequently used a nearby circle that depicted a plan view of a waste basket. When they were asked how one might represent a filing cabinet on the map, children generated a series of strictly iconic shapes and laughed uproariously at the experimenter’s suggestion that an asterisk could be used instead. We believe that these errors and misunderstandings occur because the child interprets the meaning of specific elements on the basis of the graphic characteristics

184

Lynn S.Liben and Roger M.Downs

of that element, without being constrained by the symbolic context of the representation as a whole. This loss of context can be abrupt. In interpreting the Chicago aerial photograph, for example, one child who had successfully identified a road, immediately pointed to a nearby semicircular grassy area and asserted that it was “cheese.” For young children, the struggle to understand geometrical correspondences poses even more problems. In reading maps and aerial photographs, young children have difficulty in maintaining scale and size relationships and thus do not consistently appreciate the essential map property of dimensional systematicity. On the Chicago aerial photograph, one child could find buildings, knew that they were buildings despite the fact that they were small, and yet claimed that a road could not be a road because it was “too skinny for two cars to fit on.” Another child correctly identified a baseball diamond, and yet claimed to be able to see the bench and the coach. Another correctly identified Lake Michigan as a lake, but proceeded to identify the boats on it as “fish.” On the Pennsylvania road map, one child picked out the Susquehanna River as a road but rejected road symbols as roads because only the former was “fat enough for two cars to go on.” Problems with a consistent understanding of perspective (viewing angle and viewing azimuth) are also apparent. On the vertical Chicago aerial photograph, a child correctly identified a building but claimed to see its doors and windows. Rows of railroad box cars aligned in parallel were identified as “book shelves.” Lake Michigan was identified as “the sky,” a statement that was justified by pointing to “the stars” (which were, of course, boats). Mountains on a molded plastic relief map were rejected as mountains because “they aren’t high enough.” The discussion of geometrical correspondences has focused on children’s comprehension of place representations, but the struggle to come to grips with geometrical correspondences is equally apparent in map production tasks. First- and second-grade children (aged 6 to 8 years) were asked to draw (and later select) what their school building would look like to a bird flying overhead. Although some children produced generic buildings, others successfully depicted the particular building. Even among particularized depictions, striking individual differences were found with respect to whether children produced plan or elevation views. Wolf and Gardner (1985) and Perry and Wolf (1986) showed how the distinction between maps and drawings emerges gradually between 5 and 8 years of age, Viewing angle is one distinguishing characteristic but, as our data show, maps are not necessarily associated with plan views despite the request for a bird’s-eye view. That such a request was at least heard is apparent from the birds that were included in the children’s drawings, as illustrated by one first-grader’s response shown in Fig. 2.

Development of Map Concepts in Children

185

Fig. 2. Firsf-grader’sbird’s-eye view of his school.

Age-linked differences were found in the choice of viewing angle. The incidence of elevation views decreased (from 32% in first grade to 8% in second grade), and plan views increased (from 27 to 44%). The percentages of mixed views remained approximately constant (8% in both grades), as did ambiguous responses (35 versus 40%). The number of children who depicted the correct shape, as seen from overhead, increased from 18 to 31%. As expected, performance on the selection task was superior to that on the production task. Another means of studying children’s understanding of geometrical correspondences was through the use of classroom maps such as the one shown in Fig. 3. The fact that the children had spent most of a school year in the classroom is important. Many researchers have explored children’s map understanding in unfamiliar experimental spaces. Our strategy allows for an estimate of mapping performance in a context that maximizes the child’s chances of success. At the same time, however, the use of the classroom map may be problematical in that it is not a prototypical map form: Fewer than 50% of children or adults accepted a classroom plan as a map on our map categorization task. These data suggest that it may be an atypical stimulus form from which to draw conclusions about children’s responses to maps. Despite this possible limitation, children did readily accept the room plan as a representation of their classroom and could identify familiar objects on it (such as windows and desks). Thus, the classroom mapping tasks permit us to examine a number of questions. Of particular interest were tasks that tapped children’s understanding of the alignment relationship between self,

186

Lynn S. Liben and Roger M. Downs

Fig. 3. Example of a classroom map used in MAPPS.

map, and space when the person is in the depicted space. This relationship requires an understanding of projective space-that is, of point of view (Piaget 8z Inhelder, 1948/1956, Part 11). To the extent that a child understands only topological properties of space, the self-map-space relationship is problematic. Children used the classroom map under aligned and unaligned conditions. In the aligned condition, the map was oriented in the same way as the classroom; in the unaligned condition, the map was rotated 180". Tasks included locating familiar people and objects on the map; locating the positions of objects or people not usually in the room; indicating directions in the classroom; and marking routes on the map to depict paths walked by one of the experimenters. We present here only the results from the self-location and the personlocation tasks (see Liben & Downs, 1986, for the other findings). Children

Development of Map Concepts in Children

187

placed small stickers on the map to indicate their own location and the location of another person who stood at various positions in the room. In the aligned condition, the percentages correct for the self-location task were 84, 95, and 92% for kindergarten, first-, and second-grade, respectively. Percentages correct (by grade) for the person-location task were 40, 63, and 83%. In the unaligned condition, percentages correct for the self-location task were 40, 70, and 8l%, and for the person-location task, 22, 55, and 66%. These findings are not surprising, given the existing literature and our theoretical framework. Judgments about the child’s own location were very accurate and better than the judgments about another person’s location; performance improved markedly with increasing age; and unaligned performance was much worse than aligned performance, particularly for kindergartners. However, the range of variation in levels of performance within grades is striking. Even in the aligned condition, some children in each grade were correct on none of the six person locations, and some children were correct on all. We used classroom maps to probe the child’s understanding of the geometrical correspondence between a highly familiar space in which the child is located, and a map of that space. We also assessed the child’s understanding of geometrical correspondences between two different representations of the same space. The child must understand that both representations have been transformed from the same portion of the Earth’s surface in ways that are dimensionally systematic but result in representations that are generalized, reduced, and symbolized differently. We developed a series of tasks to explore this aspect of map understanding. In one task, children transferred locations from a three-dimensional relief model of a familiar local area to a contour map of the same area. They were shown a 16 x 16 x 2:’’ three-dimensional, cardboard relief model that represented the local area of approximately 12 x 12 miles. Each layer represented a 100’ contour increment, with terrain ranging from approximately 600 to 2100’ above sea level. Seven colored flags were placed on the model, in locations varying with respect to how fully defined they were by landscape features, and thus, with respect to how readily their locations could be described topologically. For example, a flag on the top of one of the mountains was considered a highly defined location because it was the only mountain of that shape and was removed from other mountains. In contrast, flags on the large, flat areas of the model were at minimally defined locations. The lack of landmarks near these flags necessitated reliance either on estimates of direction or on distance relative to the surrounding terrain. Each child was given a board to which a 7 x 7’’ contour map of the same area had been taped. (Previous lessons had introduced contour maps in general, and this contour map in particular.) The model was placed on a table in the classroom. Children were given rectangular stickers, colored identically

188

Lynn S. Liben and Roger M.Downs

to the seven flags on the model. They were asked to put the stickers on their map to show the location of the flags on the model. Given that children were surrounding the model and moving about freely, no attempt was made to control alignment between map and model. Only a casual suggestion concerning alignment was made at the beginning: Children were told that they might find it helpful to make the map go the same way as the model. Collapsing over children and locations, performance was poor, averaging 27% correct in first grade and 40% correct in second grade. The expected differences among locations were evident (see Fig. 4). In the second grade, for example, performance was, as predicted, highest (59%) on the mountain with a distinctive shape. Performance on another mountain peak was poor (30Vo), because similarly shaped peaks were nearby. The percentage correct on the locations in the open areas sank to as low as 22%, reflecting the number of confusable areas, and the need for metric and regional accuracy in specifying the correct location. As the relatively good performance (58%) in the second grade for the location in the corner indicates, accuracy in an open area is possible if cues are available (in this case the edges of the model and the map). Again, striking variations were evident within grades, not only with respect to the number of correct responses, but also with respect to individual children’s strategies. Some children appeared oblivious to the need to consider and find isomorphisms between model and contour map. They arrayed stickers in the same spatial dispersion as the flags on the model, but they

Fig. 4. Flag locations: percentage correct in first and second grades.

Development of Map Concepts in Children

189

showed no observable concern for the relative alignment between model and map. Many of these children produced remarkably accurate responses in the sense of the relative metric relationships among the set of seven stickers. They were scaled correctly, in that the child had accommodated the scale reduction from the 16 x 16” model to the 7 x 7” map. These arrangements were totally inaccurate with respect to locations, however, because the projective spatial relationships were ignored. Other children appeared to appreciate the need to establish isomorphisms, carefully aligned the map and the model, and proceeded to array the stickers correctly. Perhaps most remarkable of all were those children who left the map in whatever haphazard relationship to the model they found it, but nevertheless located most or all of the stickers at the correct locations, with no overt evidence of cognitive struggle. Based on our empirical work, a more complex picture of the young child’s map understanding emerges than that represented in the conventional wisdom. Although some abilities may appear early, and some specific applications may appear to be easy, successes are limited to a restricted set of conditions. In the context of a highly familiar space, children as young as kindergartners are skilled in understanding the holistic correspondence between that space and a map. However, the ability to recognize and/or act on isomorphisms between map and space decreases (1) as the alignment between map and space is reduced (performance was far worse in unaligned conditions); (2) as potentially confusable alternatives are increasingly available (performance was worse for locating objects on furniture symbolized by a shape that was highly similar to another nearby shape than for highly differentiable symbols at discrete locations); and (3) as the need for metric (as opposed to topological or projective) solutions increases (performance was worse for locations in undifferentiated areas than in areas offering nearby landmarks). When dealing with two different representations of the same place, performance was generally poor for all grades tested. These tasks taxed most children’s ability to understand maps. Within a particular map task, levels of success varied by item. Where confusable alternatives were available, performance was poor; where locations were near salient landmarks or features (e.g., corners of the contour map), performance was better. This suggests, in turn, the difficulty of finding isomorphisms between alternative representations of the same space. Even if the child knows that the representations show the same place (the holistic stand-for relationship is given), matching the two representations depends upon relating specific elements of each (the componential stand-for relationship). The process of matching is complex and demanding. As indicated by data from the task requiring matching of contour map to relief model, children may be successful in some but not all parts of the matching process. Those children who were able to deal with the scale reduction and the dimensional reduction had mastered two of the

190

Lynn S. Liben and Roger M. Downs

transformations necessary for establishing geometrical correspondences, but they failed to carry out the alignment transformation at the same time. Those children who were successful on the task without physically aligning the map and the model demonstrated that geometrical correspondences could be executed cognitively. The foregoing findings are only a sample of the data we have collected. Although neither the data collection nor the description here is exhaustive, they allow us to draw some conclusions about the development of map understanding in young children in the section that follows.

V. Map Understanding in Young Children Revisited Despite our objection to the conventional wisdom that map understanding is early and easy, we would not replace this characterization with an equally simplistic opposite: late and difficult. At a very early age, perhaps by the age of 3, children have begun to form a rudimentary concept of a map. Strikingly, by the age of 5 or 6 years, without explicit instruction and with only minimal and incidental exposure to varied map forms, children have formed definite opinions about forms of place representations that fall within and outside the map category. Although children of a given age disagree or are uncertain about some place representations, many of these same representations are of equally questionable status for adults. In this respect, children and adults share similar beliefs about what constitutes a map. They do, however, differ insofar as the map category broadens with age. Whether the driving force behind this expansion is a function of exposure, instruction, or changing developmental level, the graphic form of the place representation clearly has a major effect on the ease or difficulty of assignment to the map category. And the particular graphic form is, in turn, a result of the combination of representational and geometric correspondences that give the map its character. Our work on the child’s understanding of representational and geometrical correspondences points to a need for caution. At one level, our data are consistent with Blaut and Stea’s (1971) view of what we have called “the competent preschooler.” There is indeed a high level of familiarity with one kind of place representation (the prototypical road map) and an appreciation of the general content of even an unfamiliar form of place representation (the aerial photograph). Children do seem to be good at recognizing the meaning of these representations in a general sense: They can identify aerial photographs and road maps as showing places (the holistic stand-for relationship); they can identify an element or two within them (the componential stand-for relationship); and, in the case of the road map, they know that maps are used for

Development of Map Concepts in Childreti

191

wayfinding. Presumably, this level of understanding has been the source of the conclusion that preschoolers can understand maps and aerial photographs. At a deeper level, however, the child’s success at spontaneously identifying one or two items on a map or an aerial photograph should not be taken as evidence that the child has understood the relationship between the representation and the referent space. Verbal facility should not be mistaken for comprehension and understanding. The use of simple, appropriate verbal labels should not be taken as necessarily indicative of deeper levels of understanding. The temptation to draw global inferences about the extent and depth of the child’s map understanding must be tempered. What exactly does possession of a map concept imply? What are we to conclude when a child assertively states, “That’s a map!” and we agree with that judgment? What does this child know about the form and function of maps? To what extent does the understanding involved in possessing a map concept go beyond the simple recognition of exemplars of a category? After extensive probing, we found that preschoolers who have rudimentary map concepts show precisely the kinds of problems with map comprehension that one would predict from Piagetian theory: (1) Their confusions of scale may be attributed to a lack of understanding of proportionality and metrics; (2) their confusions of perspective may be attributed to immature concepts of projective space; (3) their failure to understand the hierarchical relationships between states and nations stem from a failure to understand class inclusion; and so on. As children master these logical operations during the concrete operational period, they are better able to deal with a variety of map comprehension and map production tasks. Virtually all children showed an appreciation for the wayfinding function of maps (and some even recognized this function for the aerial photograph), but only one child (a 6-year-old) voiced any other function for either representation. Lest that characterization be taken to apply only to preschoolers, most adults (and researchers) share this limited view of the function of maps. Where adults and preschoolers surely differ, however, is in the ability to carry out the wayfinding function with the aid of a map. Despite the knowledge that maps were for wayfinding, only a few children were able to pick out and follow a route on the road map. Bearing in mind this caveat about the interpretation of data, let us return to our characterization of map understanding in young children. The comment about the use of the road map for wayfinding would appear to conflict with studies (e.g., Bluestein & Acredolo, 1979; Presson, 1982) demonstrating that even children in the age range from 2; to 3 years can use maps for wayfinding in room-sized spaces. Although we do not question these other studies, we would stress the importance of specifying precisely what the child is and is not able to do under

192

Lynn S Liben and Roger M. Downs

what circumstances and with what materials. We found that, when the mapping tasks and materials were simple, even very young children do show a facility with maps, an understanding of what they represent, and how they may be used. Thus, when mapping tasks (1) are devoid of significant memorial demands; (2) avoid problems of determining relationships among self, map, and space; and (3) minimize the need for metric accuracy, even kindergarten children achieve success at using maps for representation. At the same time, these accomplishments have distinct limits. We found that many children showed a dramatic drop in performance when, for example, the map and the space were not aligned. Furthermore, tasks that required children to move between three- and two-dimensional representations or between various forms of two-dimensional representations of the same space met with mixed levels of success. Although some tasks were almost universally failed by kindergarten through second-grade children (e.g., being able to relate the viewing azimuth on an aerial photograph to a map of the same area), and others were almost universally mastered (e.g., being able to mark on a map one’s own location and the direction that one is facing), the pattern emerging from our data follows an anticipated age-linked course, with performance increasing over the three grade levels studied. Apart from the expected improvement in map performance with age, however, even more striking is the range of performance within grades. Individual differences were evident not only with respect to the range of scores within each grade on many mapping tasks (i.e., with some children answering no items correctly and others answering every item correctly), but also with respect to the ways in which children approached the mapping tasks. The development of map understanding is, therefore, a gradual, multifaceted, and complex process. This issue of complexity distinguishes our interpretation from that prevalent in the literature. As an illustration of this difference, we return to the finding that many children show a dramatic drop in performance when map and space are not aligned. One might, for example, take Landau’s (1986) and Presson’s (1982) work as offering a contradictory assertion. And in one sense, they do offer an alternative characterization in which the drop in performance is not dramatic. Under certain circumstances, the effect of nonalignment on performance can be minimized. But what exactly has been shown under those circumstances? The Landau (1986) and Presson (1982) studies lead easily to unwarranted inferences. Of particular concern is the kind of map that was presented to these very young children. In many ways, the form of map is analogous to the “See Spot run” type of sentence that appears in early reading texts. It is a stripped-down, minimalist version of the map, which loses much of the essence of the map concept. “See Spot run” is a sentence, but one would not want to make exaggerated inferences about the child’s ability to read, to understand syntax and

Development of Map Concepts in Children

I93

grammar, to appreciate rhetorical construction, and so on using the comprehension of such a sentence as the evidentiary basis for an argument. We would argue that researchers who study map understanding are in danger of making just such a mistake. “See Spot run” maps proliferate in the research paradigm that is emerging. A map is a generalized, symbolic spatial representation of reality that has been transformed from the spherical surface of the Earth (or any other area) in some dimensionally systematic way. Ironically, to the extent that more complex place representations are used, one is in danger of not probing deeply enough into the level of map understanding that the child possesses. Taken together, the “See Spot run” maps and the superficial analysis of complex maps are part of the basis for the early and easy conventional wisdom about map understanding. We would argue for a more valid characterization: Map understanding indeed begins early, but it progresses through a complex and difficult sequence of developments that are simply not well understood at present. An understanding of this sequence can come only from a theoretical framework that allows us to appreciate not only the development of map understanding but also the understanding of maps.

VI. Maps as Symbolic Representations Underpinning our discussion of maps is a fundamental question: How do children know what they are looking at? When and how do children understand that a pattern of lines and colors or gray tones on a sheet of paper stands for a particular place in the real world? In other words, if the child knows that this thing (a particular representation) is an example of an X(where X i s a map, a photograph, a drawing, etc.), then the child proceeds to interpret and use it in a particular way. The child must have a concept of an X in order to understand what that type of representation shows and hence what the various components mean, taken singly and taken together as parts of a whole. That process of interpretation is built on a set of expectations (derived from the concept) about what it can show (content), how it shows that content (form), and how it can be used (function). A similar process would apply to the production of any form of symbolic representation. This fundamental question has two parts: (1) On what basis does the child know? What are the “things”-the necessary and sufficient properties, that the child uses to decide that X is a map (or any other form of symbolic representation)? This question is central to the recognition or identification of any symbolic representation. In the case of maps, it requires that we frame our research within the context of cartographic theory. As our discussion of Perry and Wolf (1986) suggests, the necessary properties of maps are neither

I94

Lynn

S. Liben and Roger M. Downs

obvious nor indisputable. (2) How, when, and under what circumstances does the child’s knowledge come into being? This part of the fundamental question is central to a cognitive-developmental approach to symbolic representation. In this respect, Perry and Wolfs (1986) work is important because it shows that children progressively differentiate one form of symbolic representation from another. Perry and Wolf focused on the distinction between the production of a map and a drawing of the same area. The process of differentiation requires an increasingly sophisticated understanding of the relationship between forms, contents, and functions of representations. Our data on the concept of a map show that the category of maps becomes broader with age and that more forms are accepted as maps. Unfortunately, the process of differentiation is not simple because the boundaries between forms of representation are not clear-cut. Map forms cross the boundaries between forms of representation (Monmonier & Schnell, 1988). For example, are orthophotoquads maps with a photographic component, or are they photographs with a map component? Are Landsat satellite images photographs of what is there, or are they carefully abstracted and computerenhanced maps that reflect best interpretations of what is really there? In both cases, the answer is important because of the expectations and beliefs that one thereby brings to bear. Berger (1980) stated a commonly held belief about photographs: It was not, however, until the 20th century and the period between the two world wars that the photograph became the dominant and most “natural” way of referring to appearances. It was then that it replaced the world as immediate testimony. It was the period when photography was thought of as being most transparent, offering direct access to the real. (p. 48)

Do either photographs or maps offer direct access to the real? If one’s answer is yes, then the process of interpretation is presumably the same as that for the real world itself, and maps are indeed transparent. The question, therefore, of how one knows what one is looking at is central to understanding representations. In effect, the child wants to get behind the particular form of the representation to its content, its meaning, to the thing that is being represented. If one knows what one is looking at (i.e., understands the form of the particular map), then one can look through it. How easy is this to do? There are conflicting approaches to understanding maps as symbolic representations. One, the early and easy approach, is tantamount to a dismissal of the question. Young children can understand maps without prior exposure and even without visual experience. The basis follows from Berger’s idea: Maps are transparent. One can see through a map immediately to the real world that it captures. If maps are transparent, then their interpretation

Development of Map Concepts in Children

195

and production may legitimately be expected to come early and easily, and the basis for the acceptance of maps as representations is obvious. We prefer an alternative approach to understanding maps as representations, one that emphasizes the idea of symbols. Maps are accepted as representations gradually as the result of an extended process of learning to see the world through maps. Far from being transparent, maps have surface properties that may render them opaque in ways that cannot be overcome without considerable exposure to and experience with a variety of map forms. Far from being immediate, understanding is possible only as a corollary of the general cognitive progressions enumerated in Piagetian theory. Moreover, far from being easy, map understanding involves a struggle through failures, marked by errors and confusions. If one does not fully understand the idea of content and especially the idea of form, then one can make mistakes (as children did in interpreting the Chicago aerial photograph and the Pennsylvania road map). In this view, part of the child‘s difficulty lies in the extent to which elements of a particular form of representation are iconic versus symbolic. As our data show, if the child believes that the elements of the representation look like things in the real world, then the child may make errors of interpretation. This is the stand-for relationship at the componential level. Thus, red roads on the map would be red in the world at large; the yellow-colored symbol for a built-up area would be interpreted as either firecrackers or eggs. At the holistic level of the stand-for relationship, Beilin’s (1983) concept of iconic realism is relevant. In discussing the reactions of 3- through 5-year-olds to color photographs, he reported instances in which children appear to believe that the photographs possess properties that are unique to the objects themselves. This shows a failure to appreciate that the representation stands for the object. The representation is interpreted as the same as the object. The balance between iconic and symbolic components within a form of representation is neither fixed nor obvious. It is a function of convention, which is, in turn, the result of decisions by producers about form and content in representations. Thus, for example, water is typically shown in blue on a map. But water need not be shown in blue (as in the infamous New York subway map of the late 1970s, in which a pale brown color was used for this purpose). And blue need not stand for water (as in the Washington, D.C. tourist map used in the MAPPS interviews, in which a shade of blue represented land). That “typically” does not mean “necessarily” is precisely the idea that people (adults and children alike) have trouble accepting and understanding. Unfortunately, however, Wohlwill’s (1973, p. 167) memorable phrase, “the Gospel according to Rand McNally,” captures a commonly held view about the nature of maps, specifying form, content, and function at one and the same time. The point is not that the “Gospel” is wrong as such;

196

Lynn S.Liben and Roger M.Downs

instead, one must learn that it takes its place alongside the Gospel according to Landsat images, according to the U.S. Geological Service, according to the local Chamber of Commerce, and so on. That such a broadening of the map concept does occur is shown by our data. Although children tended to reject photographic place representations as maps, adults were increasingly willing to include some photographic forms as maps. In order to understand the interaction between convention and form of symbolic representation, we must confront another set of issues: the roles of creativity and change in modes of representation. Map forms have changed continuously, and they will presumably continue to change. A problem occurs when the set of expectations (the Gospel according to Rand McNally) is treated as static and frozen, as the way that it is and has to be. Again, developmental evidence indicates that this is not the case. Wood (1977) explored the parallels between means of landform and relief representation (hiI1-form symbols) throughout the history of cartography and within the development of children. He identified three ordered sequences of change in both contexts: a pictureto-abstraction shift, a profile-to-plan shift, and a generic-to-unique shift. The result of the three sequences is the prototypical map form of an abstract, plan view of a specific part of the Earth’s surface, using conventional symbols. Thus, just as there are ordered changes in the ways in which individuals map their worlds with development, so too, there are ordered changes in the ways that societies generate cartographic conventions. Neither children nor maps are static.

VII. Summary We expect that many readers encountered this article with the beliefs that maps are highly specialized devices primarily used for wayfinding; that they represent the spatial world in a single, correct form; that they are readily transparent; and that their sole contribution to psychology is their role in externalizing environmental cognition. By discussing the myriad functions and forms of maps, by highlighting their symbolic nature, and by considering some of the misconceptions about maps, we have attempted to demonstrate the value of maps for addressing a wide range of developmental questions. Our review of past research literature suggests that research conducted within individual disciplines has both strengths and limitations. Work in the psychological tradition is characterized by attention to important subject characteristics and to carefully described and implemented research designs, procedures, coding, and analyses. At the same time, this work reveals, at best, highly restricted views about maps, and at worst, fundamental misconceptions about maps. Work in the geographic and environmental traditions, in

Development of Map Concepts in Children

197

contrast, samples a broader range of map forms and functions, but it suffers from inattention to procedural details that makes the conclusions less compelling than they might otherwise be. A conventional wisdom is emerging from the work in both traditions: That children’s map understanding occurs extremely early and extremely easily. The limitations of both research traditions, however, suggest the need for caution in accepting this view. Developmental and cartographic theories provide a compelling reason to reexamine the early and easy view and suggest the need for alternative conceptual and empirical approaches. We have argued that future work should integrate the traditions of psychology and geography. Illustrative data from an interdisciplinary program of research were presented. We described work demonstrating the gradual and difficult process of mastering the representational and geometric correspondences that link the map to its referent in the world. Our data suggest that there are significant achievements in map conceptualization (the understanding of the concept of a map), map identification (understanding the formal components of a map), and map utilization (the ability to use maps). Our data support the view that maps are not transparent and that children’s abilities to understand, use, and create maps are linked to their developing representational and spatial skills. In concluding, we should acknowledge that we have deliberately pushed interpretations about understanding maps as symbolic representations to the extreme. The reason for this strategy is simple: We believe that work on maps-both in the public schools and in academia-is assumed to be an expendable and irrelevant luxury. In large measure, this condition stems from the erroneous belief that maps are transparent miniaturizations of the world, and from the correlative failure to appreciate the symbolic nature and power of maps. To the extent that educators and researchers come to share our view, perhaps future generations of children can be spared from equating maps with locations of state capitals, and perhaps future cohorts of psychologists can be dissuaded from reflexively skipping an article with the word map in its title. ACKNOWLEDGMENTS We express our deep appreciation to a number of our colleagues-Linda Acredolo, Gary Allen, Harry Beilin, Debra Daggs, Norman Freeman, Jeffrey Lockman, and Nora Newcombe-who read an earlier draft of this article and provided a variety of insightful and thought-provoking comments. We likewise acknowledge with thanks the enthusiastic cooperation of the participating administrators, teachers, parents, and children at the Bellefonte Elementary School, Bellefonte, Pennsylvania, the Jewish Community Center of State College, Pennsylvania, and the Child Development Laboratory at Penn State; the always cheerful and efficient work by Sandy Ranio

198

Lynn S. Liben and Roger M. Downs

in preparing project materials; and the dedicated help in data collection and coding of Steven Knepper, Corinne Lewkowicz, Ann Merriwether, Robin Yaure, and Candice Yekel. We appreciate permission to reproduce materials from previous publications. In particular, we thank United Features Syndicate, Inc. for permission to reproduce the Peanuts cartoon from November 14, 1975; The Hogarth Press, Harper & Row, and Mrs. Laura Hwley for permission to reproduce the quotation from Point Counter Point by Aldous Huxley; and The Society for Research in Child Development and Linda Acredolo for permission to reproduce a figure from Child Development. We are also grateful to the National Institute of Education for its financial support (Grant #NIE-G-83-0025) for much of the research reported here. (Any opinions, findings, and conclusions expressed are those of the authors and do not necessarily reflect the views of the Institute or the U.S. Department of Education.) In addition, we thank the Office of Research and Graduate Studies of the College of the Liberal Arts at Penn State for funds for the preparation of this article. This article is part of an ongoing, completely collaborative effort by the two authors, and thus order of authorship is arbitrary.

REFERENCES Abel, R. R., & Kulhavy, R. W. (1986). Maps, mode of text presentation, and children’s prose learning. American Educational Research Journal, 23, 263-274. Acredolo, L. P. (1981). Small- and large-scale spatial concepts in infancy and childhood. In L. S. Liben, A. H. Patterson, & N. Newcombe (Eds.), Spatial representation and behavior across the life span: Theory and application (pp. 63-81). New York: Academic Press. Beilin, H. (1983). Development of photogenic comprehension. Art Education. 36, 28-33. Berger, J. (1980). About looking. New York: Pantheon. Bertin, J. (1983). Semiology of graphics: Diagrams, networks, maps. Madison: University of Wisconsin Press. Blades, M., & Spencer, C. (1986a). The implications of psychological theory and methodology for cognitive cartography. Cartographica, 23, 1-13. Blades, M., & Spencer, C. (1986b). Map use by young children. Geography, 71, 47-52. M. (1969). Space, structure, and maps. Place Perception Research Reports, 2, (6). Blaut, .I. Blaut, J. M., McCleary, G. S., Jr., & Blaut, A. S. (1970). Environmental mapping in young children. Environment and Behavior, 2, 335-349. Blaut, J. M., & Stea, D. (1971). Studies in geographic learning. Annals of the Association of American Geographers, 61, 387-393. Bluestein, N., & Acredolo, L. (1979). Developmental changes in map-reading skills. Child Development, 50, 691-697. Brown, A. L. (1989). Analogical learning and transfer: What develops? In S. Vosniadou & A. Ortony (Eds.), Similarity and analogical reasoning. New York Cambridge University Press. Bruner, J. S. (1959). Learning and thinking. Harvard Educational Review, 29, 184-192. Cohen, R. (Ed.). (1985). The development of spatial cognition. Hillsdale, NJ: Erlbaum. Debache, J. S. (1987). Rapid change in the symbolic functioning of very young children. Science, 238, 1556-1557. Dodgson, C . (1939). The complete works of Lewis Carroll. London: Nonesuch Press. Downs, R. M. (1981). Maps and mappings as metaphors for spatial representation. In L. S. Liben, A. H. Patterson, & N. Newcombe (Eds.), Spatial representation and behavior across the life span: Theory and application (pp. 143-166). New York Academic Press.

Development of Map Concepts in Children

I99

Downs, R. M. (1985). The representation of space: Its development in children and in cartography. In R. Cohen (Ed.), The development of spatialcognition @p. 323-345). Hillsdale, NJ: Erlbaum. Downs, R. M., & Liben, L. S. (1988). Through a map darkly: Understanding maps as representations. Genetic Epistemologist, 16, 11-18. Downs, R. M., Liben, L. S., & Daggs, D. G. (1988). On education and geographers: The role of cognitive developmental theory in geographic education. Annals of the Assocation of American Geographers, 78, 680-700. Eastman, J. R. (1985). Graphic organisation and memory structures for map learning. Cartographica, 22, 1-20. Ekman, G., Lindman, R., & William-Olsson, W. (1961). A psychophysical study of cartographic symbols. Perceptual and Motor Skills, 13, 355-368. Feldman, A., & Acredolo, L. P. (1979). The effect of active versus passive exploration on memory for spatial location in children. Child Development, 50, 698-704. Feldman, D. H. (1980). Beyond universals in cognitive development. Norwood, NJ: Ablex. Gardner, H., & Wolf, D. (1987). The traditional views of exploration and the cognitive revolution. In D. Gorlitz, and J. F. Wohlwill (Eds.), Curiosity, imagination, andplay (pp. 306-325). Hillsdale, NJ: Erlbaum. Gauvain, M., & Rogoff, B. (1986). Influence of the goal on children’s exploration and memory of large-scale space. Developmental Psychology, 22, 72-77. Hart, R. (1979). Children’s experience of place. New York: Irvington. Harvey, P. D. A. (1980). The history of topographical maps: Symbols, pictures and surveys. London: Thames & Hudson. Hazen, N. L., Lockman, J. J., & Pick, H. L. (1978). The development of children’s representations of large-scale environments. Child Development, 49, 623-636. Hirtle, S. C., & Mascolo, M. F. (1986). Effect of semantic clustering on the memory of spatial locations. Journal of Experimental Psychology: L.earning Memory, and Cognition, 12, 182-189. Huxley, A. (1928). Point counter point. London: Chatto & Windus. Kosslyn, S. M., Ball, T.M., & Reiser, 9. J. (1978). Visual images preserve metric spatial information: Evidence from studies of image scanning. Journal of Experimental Psychology: Human Perception and Performance, 4, 47-60. Landau, B. (1986). Early map use as an unlearned ability. Cognition, 22, 201-223. Landau, B. (1988). Spatial knowledge in blind and sighted children. In J. Stiles-Davis, M. Kritchevsky, & U. Bellugi (Eds.), Spatial cognition: Brain bases and development (pp. 343-371). Hillsdale, NJ: Erlbaum. Landau, B., & Gleitman, L. R. (1985). Language and experience: Evidence from the blind child. Cambridge, MA: Harvard University Press. Landau, B., Gleitman, H., & Spelke, E. (1981). Spatial knowledge and geometric knowledge in a child blind from birth. Science, 213, 1275-1278. Landau, B., & Spelke, E. (1985). Spatial knowledge and its manifestations. In H. M. Wellman (Ed.), Children’s searching (pp. 27-52). Hillsdale, NJ: Erlbaum. Landau, B., Spelke, E., & Gleitman, H. (1984). Spatial knowledge in a young blind child. Cognition, 16, 225-260. Levine, M., Jankovic, 1. N., & Palij, M. (1982). Principles of spatial problem solving. Journal of Experimental Psychology: General, 111, 157-175. Liben, L. S. (1981). Spatial representation and behavior: Multiple perspectives. In L. S. Liben, A. H. Patterson, & N. Newcombe (Eds.), Spafial representation and behavior across the life span: Theory and application (pp. 3-36). New York: Academic Press. Liben, L. S. (1988). Conceptual issues in the development of spatial cognition. In J. Stiles-Davis, M. Kritchevsky, & U. Bellugi (Eds.), Spatial cognition: Brain bases and development (pp. 167-194). Hillsdale, NJ: Erlbaum.

200

Lynn S. Liben and Roger M. Downs

Liben, L. S., & Downs, R. M. (1986). Children'sproductionand comprehension of mapx Increasing graphic literacy. Final Report to the National Institute of Education, No. G-83-0025. Liben, L. S., & Newcombe, N. (1988). The appearance and disappearance of barrier effects in children and adults. Unpublished manuscript, The Pennsylvania State University, University Park. Liben, L. S., Patterson, A. H., & Newcombe, N. (Eds.). (1981). Spatialrepreentation and behavior across the life span: Theory and application. New York: Academic Press. Lobeck, A. K. (1956). Things maps don't tell us. New York: Macmillan. Lockman, J. J., &Pick, H. L., Jr. (1984). Problems of scale in spatial development. In C. Sophian (Ed.), Origins of cognitive skills (pp. 3-26). Hillsdale, NJ: Erlbaum. Mandler, J. M. (1988). The development of spatial cognition: On topological and euclidean representation. In J. Stiles-Davis, M. Kritchevsky, & U. Bellugi (Eds.), Spatial cognition: Brain bases and development (pp. 423-432). Hillsdale, NJ: Erlbaum. McNamara, T. P., Ratcliff, R., & McKoon, G. (1984). The mental representation of knowledge acquired from maps. Journal of Experimental Psychology: Learning, Memory, and Cognition, 10, 723-732. Millar, S. (1988). Models of sensory deprivation: The nature/nurture dichotomy and spatial representation in the blind. International Journal of Behavioral Development, 11, 69-87. Monmonier, M. S. (1985). Technological transition in cartography. Madison: University of Wisconsin Press. Monmonier, M. S., & Schnell, G. A. (1988). Map appreciation. Englewood Cliffs, NJ: Prentice-Hall. Perry, M. D., & Wolf, D. P. (1986, May). Mapping symbolic development. Paper presented at the Sixteenth Annual Symposium of the Jean Piaget Society, Philadelphia, PA. Piaget, J. (1965). The child's conception of number. New York: Norton. (Original work published 1941 as La gdnese du nombre chez l'erlfant) Piaget, J., & Inhelder, B. (1956). The child's conception of space. New York: Norton. (Original work published 1948 as La reprksentation de I'espace chez l'enfant) Pick, H. L., Jr. (1987). Information and the effects of early perceptual experience. In N. Eisenberg (Ed.), Contemporary topics in developmental psychology. New York: Wiley. Presson, C. C. (1982). The development of map-reading skills. Child Development, 53, 196-199. Robinson, A. H., Sale, R. D., Morrison, J. L., & Muehrcke, P. C. (1984). Elements of cartography. New York: Wiley. Scholnick, E. K., Frank, R. E., Fein, G. G., & Schwartz, S. (1986, May). Routes to representation. Paper presented at the Sixteenth Annual Symposium of the Jean Piaget Society, Philadelphia, PA. Shepard, R. N., & Hurwitz, S. (1984). Upward direction, mental rotation, and discrimination of left and right turns in maps. Cognition, 18, 161-193. Siegel, A. W., & Schadler, M. (1977). Young children's cognitive maps of their classroom. Child Development, 48, 388-394. Smith, C. D. (1987). Cartography in the prehistoric period in the Old World: Europe, the Middle East, and North Africa. In J. B. Harley & D. Woodward (Eds.), The history of cartography: Vol. I. Cartography in prehistoric. ancient, and medieval Europe and the Mediterranean (pp. 54-101). Chicago, IL: University of Chicago Press. Southworth, M., & Southworth, S. (1982). Maps. Boston, MA: Little, Brown. Spencer, C., & Darvizeh, Z. (1981). The case for developing a cognitive environmental psychology that does not underestimate the abilities of young children. Journal of Environmental Psychology, 1, 21-31. Spencer, C., Harrison, N., & Darvizeh, 2. (1980). The development of iconic mapping ability in young children. International Journal of Early Childhood, 12, 57-64.

Development of Map Concepts in Children

201

Stea, D. (c. 1972). The use of environmental modelling (“toy play”)for studying the environmental cognition of children and adults (mimeograph). Source unknown. Stea, D., & Blaut, J. M. (c. 1970). Some preliminary observations on spatial learning in Puerto Ricun school children (mimeograph). Source unknown. Stea, D., & Blaut, J. M. (1973). Notes toward a developmental theory of spatial learning. In R. M. Downs & D. Stea (Eds.), Image and environment (pp. 51-62). Chicago, IL: Aldine. ’Ikeib, M. (1980). Mapping experience. Design Quarrerly, 115 (Whole Issue). Uttal, D. H., & Wellman, A. M. (1987, April). Young children’s representation of information acquired from maps. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Wellman, H. M. (Ed.). (1985). Children’s searching. Hillsdale, NJ: Erlbaum. Wohlwill, J. F. (1973). The environment is not in the head! In W. F. E. Preiser (Ed.), Environmental design m e a r k Vol. two. Symposia and workshops (pp. 166-181). Stroudsburg, PA: Dowden, Hutchinson, & Ross. Wolf, D., & Gardner, H. (1985). Broadening literacy: A final report to the Carnegie Corporation. Cambridge, MA: Harvard Graduate School of Education. Wood, D. (1977). Now and then: Comparisons of ordinary Americans’ symbol conventions with those of past cartographers. Prologue, 151-161.

This Page Intentionally Left Blank

THE DEVELOPMENT OF SPATIAL PERSPECTIVE TAKING

Nora Newcombe DEPARTMENT OF PSYCHOLOGY TEMPLE UNIVERSITY PHILADELPHIA, PENNSYLVANIA 19122

1. INTRODUCTION

11. WHAT DOES PERSPECTIVE TAKING ASSESS? A. EGOCENTRISM B. RULES FOR SEEING C. OPERATIVE DEVELOPMENT 111. HOW CAN SPATIAL LOCATION BE ENCODED? A. TOPOLOGICAL, PROJECTIVE, AND EUCLIDEAN SPACE B. CRITICISMS C. ANOTHER MODEL O F LOCATION CODING IV. FACTORS AFFECTING SUCCESS ON PERSPECTIVE-TAKING TASKS A. ATTRIBUTES OF SUBJECT B. ATTRIBUTES OF TASK C. ATTRIBUTES OF DISPLAY D. RESPONSE MODE E. SUMMARY V. RELATED TASKS A. EUCLIDEAN SPACE B. MENTAL ROTATION VI. CONCLUSION REFERENCES

I. Introduction Piaget and his colleagues invented a multitude of tasks designed both to engage the interest of children and to reveal to curious adults the nature of children’s thought. One of the most famous is the “Three Mountains 203 ADVANCES 1N C H l L D DEVELOPMENT AND BEHAVIOR, VOL. 22

Copyright 6 1989 by Academic Press, Inc. All rights of reproduction in any form reserved.

204

Nora Newcombe

Problem,” in which children are asked to look at a display of three mountains and to indicate what view would be seen by an observer in another position, Piaget and Inhelder (1948/1967) reported that children made many errors on this problem, up to the age of 9 or 10 years. In particular, when children failed to indicate the observer’s true view, they often picked their own view instead. What does this pattern of performance tell us about children’s thought? Both Piaget and Inhelder and subsequent authors, have given a variety of answers to this question. The perspective-taking problem has been most popularly believed to index a global characteristic called “egocentrism,” said to be exhibited in many other cognitive tasks, as well as in social understanding. Other writers, however, have regarded the three-mountains task as us about a more restricted part of cognitive development-namely, how children remember the location of objects and what they do with this information. In this article, I argue that Piaget himself made primarily the latter claim. Section I1 of this article is addressed to the issue of whether egocentrism is a useful concept, and it contains a review of evidenceconcerning what young children do understand about “rules for seeing.” The section also includes a discussion of the nature of the relationship between development of spatial perspective-taking skills and operative development, in Piaget’s theory. The essential argument of Section I1 is that the most interesting way to approach spatial perspective taking is to focus primarily on what it can tell us about the encoding and transformation of information about spatial location. Thus, Section 111 deals with proposals concerning the nature of spatial encoding. Piaget and Inhelder discussed three possible systems for location coding: topological, projective, and Euclidean space. The first was said to precede the second two in development. Criticisms of Piaget and Inhelder’s proposal, and alternative ideas about spatial coding proposed by Huttenlocher and Newcombe (1984) are also described in this section. The longest section of this article is Section IV, a review of the spatial perspective-taking literature, which at this point includes about 100 studies. One purpose of the review is to update an earlier one by Fehr (1978). In addition, however, Fehr aimed essentially to catalogue the variables that influence whether perspective taking, considered as a single variable, appears earlier or later than Piaget claimed. Such a purpose has preoccupied many of the investigators of perspective-taking tasks, but the focus makes sense only on the assumption that perspective taking is a unitary ability, of which an individual can possess more or less. In this article, perspective-taking tasks are regarded as multiply determined. Thus, performance reflects all of the abilities required to perform the task and is revealing only within the context of analysis of the task’s requirements. The purpose of Section IV is to anaIyze

Development of Spatial Perspective Taking

205

the components of perspective-taking tasks, in order to clarify why some tasks are harder than others. Perspective-taking tasks are not, of course, the only ones that have been used to examine spatial development. Section V considers the most closely related other tasks: (1) allocentric placement, (2) understanding of horizontality and verticality, and (3) mental rotation. The purpose is to examine whether the picture of spatial development gained from perspective-taking problems is similar to that resulting from work with these other paradigms. Section VI summarizes the conclusions of this article. Specifically, 1 argue that at least by the age of 5 years, children encode the location of small movable objects with respect to a framework of external landmarks, and that this type of encoding continues into adulthood. Some evidence indicates that children even as young as 3 years of age may encode location in the same way, and it follows that children this young may be expected to succeed at perspective-taking tasks that most naturally involve such a representation.

11. What Does Perspective Taking Assess? A. EGOCENTRISM

I. Is Egocentrism a Tfait? When children are shown a three-mountains display and are asked about the view of another observer, they pick their own point of view as being that of the other person. The natural inference seems to be a radical inability to identify with the experience of other people. Indeed, Piaget and Inhelder wrote of such performances that “the child fails to realize that different observers will enjoy different perspectives and seems to regard his own point of view as the only one possible” (p. 213). Children seem to show such focus on the self in other situations as well. For instance, they seem to assume in talking to other people that their partner knows everything they do (communicative egocentrism), they have difficulty guessing what other people will feel in certain situations, or imagine others will feel what they do (affectiveegocentrism),and they fail to understand that other people know different facts from those they know (cognitive egocentrism). From these separate findings many observers have concluded that through much of middle childhood children suffer from a global quality or trait of egocentrism. Such a conclusion seems to predict substantial correlations among egocentrism scores in various situations. This prediction has undergone the same vicissitudes that have befallen other hypotheses concerning the existence of

206

Nora Newcombe

traits in human personality (see, e.g., Epstein, 1979; Mischel, 1968). Several recent reviewers have evaluated whether scores on spatial perspective-taking tasks are correlated with scores on measures of egocentrism in other cognitive tasks, as well as in affective and communicative situations (Ford 1979, 1985; Rushton, Brainerd, & Pressley, 1983; Shantz, 1975; Waters & Tinsley, 1985). Ford (1979) concluded that intercorrelations among various measures of egocentrism were low and not always even significant. He proposed that egocentrism be considered only in task-specific terms. Rushton et al. (1983) acknowledged that existing correlations were low, but argued that simple psychometric problems with reliability accounted for this fact, and that each type of egocentrism needs to be assessed in multiple ways and aggregated across occasions and situations. Waters and Tinsley (1985) agreed that correlations may be kept low by problems with reliability, and added that existing work had problems with the statistical control of age in mixed-age designs and with the use of scores reflecting total errors rather than egocentric errors per se. They concluded that the hypothesis of cross-domain relationships in egocentrism has simply never been properly evaluated. Whether future research built on these methodological refinements will provide evidence for a trait-like conception of egocentrism is, of course, unknown. More interestingly for the purposes of this article, however, both sides in this debate have assumed that the existence of a trait of egocentrism is a crucial issue for Piagetian theory. Waters and Tinsley were to some extent an exception, arguing that egocentrism falls into the class of research on individual differences, a topic in which Piaget never took much interest. But the matter goes beyond this point.

2. The Place of Egocentrism in Piagetk Theory Egocentrism was regarded as an explanatory construct in Piaget’s early writings (e.g., Piaget 1923/1926), but by the time he worked on space, he had changed this position. Egocentrism had become a descriptive attribute, a characteristic of the early stages of several developmental sequences. It did not, however, serve to explain development, nor did it constitute part of a structural description of thought such that any generality could be expected of it. That is, egocentrism is not like the logic of classes, for example. Any task involving class logic ought to be related, to some extent, to any other task involving class logic, because class logic is part of a structure, is a concrete operational grouping. The same is not true of egocentrism. A better analogy for egocentrism might be another heuristic principle of development, the cephalocaudal principle. Morphological development in utero proceeds roughly in a head-to-toe fashion, and motor skills in the first year develop roughly in the same fashion. However, one would not thereby

Development of Spatial Perspective Taking

207

expect a fetus to show directed grasping. Nor would one expect even that rate of fetal development would correlate with rate of acquisition of motor skills. Admittedly, Piaget sometimes wrote as if he believed otherwise. Thus, one can find claims in The Childs Conception of Space (1967) such as “it is the egocentric illusion which prevents these children from reversing the left-right, before-behind relations and thereby rotating perspectives along with their changing viewpoints” (p. 218). Nevertheless, in the context of the whole book, such passages seem almost to be asides. Much more weight is given to the perspective task as indicative of understanding of projective space, and the overarching purpose of the book and of the large number of experiments reported in it is to explore the nature of the child’s understanding of space. Moreover, as Morss (1987) discusses at length, the evidence available to Piaget did not clearly support the egocentrism hypothesis, and indeed was at least equally compatible with the hypothesis that young children have difficulty understanding the idea of “point of view” at all. That subsequent writers have concentrated on the issue of egocentrism is understandable, but also regrettable. This concentration has had the consequence that many studies in the literature have been aimed at demonstrating simply that under some suitable set of circumstances, young children do show an awareness that other observers see things differently from the way they do. This point has now been clearly established (see Section 11,B). The more important focus in terms of spatial representation is on understanding why children do have difficulty dealing with visual perspective taking in some situations where the locations of various objects have to be computed. Information about spatial representation must come from computation tasks, not problems simply requiring recognition that others see something different from oneself.

3. Egocentrism in Infancy and Childhood One of the problems with the term egocentrism is that it has had varying definitions both conceptually and operationally. In infancy, egocentrism has been inferred from searching for an object after a change in position has occurred in a way that ignored the change in position. Some writers have concluded that infants represent location only with respect to their own bodies. Debate over this interpretation has been extensive. Presson and Somerville (1985) summarized several problems: Infants’ mistakes are at least partly due to the repetition of motor habits rather than to the use of incorrect coding schemes; and correct responding can appear as young as 6 months when landmarks are familiar or salient or when the amount or type of motion is changed (Acredolo, 1979; Acredolo & Evans 1980; Bremner, 1978; McKenzie, Day, & Ihsen, 1984; Rieser, 1979). One possibility for explaining variations in the age at which egocentric responding declines is to postulate that location is

208

Nora Newcombe

always represented with respect to some landmark, including the infant’s body, other people in the room, and stable external landmarks. These systems sometimes come into conflict with each other or with motor habits. With age, the infant sorts out which systems are more useful. Whatever the outcome of this debate about infancy, for the purposes of this article, an important point is that egocentric responding on a search task following movement has not been found at later than 2; or 3 years (Braine & Eder, 1983). In addition, a variety of studies, reviewed herein shortly, indicate that children know that observers in different vantage points have different views, beginning by the age of 2 or 3 years. Thus, egocentrism, as inferred from choices by elementary school children on perspective-taking tasks must mean something different from egocentrism as inferred from infants’ search ing. One possibility is that egocentric choices on a perspective-taking task may indicate nothing specific about spatial coding but rather simply constitute default options in situations where the child is confused (e.g., Aebli, 1967). After all, the child’s own view is at least a possible view of the display. In summary, egocentrism is not a useful term in considering perspectivetaking tasks. It is not central to Piagetian theory, with no claims being made for its explanatory value or its place in a structural theory. Furthermore, it no longer has exact meaning, except in infancy, as a hypothesis about spatial coding, and even for infancy, the status of the hypothesis is in some doubt. B. RULES FOR SEEING

A second view of what perspective taking assesses has been advanced by Flavell and his colleagues. They explicitly divided performance on perspective taking into two areas: (1) rules for knowing what another person sees in general, and (2) computational processes for deciding what that person sees in particular (Flavell, Omanson & Latham, 1978). The Flavell group has concentrated on the investigation of the former domain, that of rules, which they defined as “general relationships among observer positions and observer visual experiences, relationships that are essentially invariant across displays” (Flavell, Omanson, & Latham, 1978, p. 462). These rules are classified as belonging to either Level 1 or Level 2. Level 1 knowledge concerns inferences about what objects can be seen from another person’s viewpoint. At Level 1, children know the that and the what of other people’s visual experience, but not the nature of that experience in detail (Masangkay et al., 1974). At Level 2, children have acquired rules such as the following: one observer has only one view of an object or array (Salatas & Flavell, 1976), observers in different positions have different views (Flavell, Flavell, Green, & Wilcox, 1981; Flavell, Omanson, & Latham, 1978; Salatas & Flavell, 1976), observers in the same position have the same view (Flavell,

Development of Spatial Brspective Taking

209

Flavell et al., 1981; Flavell, Omanson, & Latham, 1978), observers in different positions have essentially the same view of homogeneous-sided objects (Flavell, Flavell, et al., 1981) objects that are farther away can be less clearly perceived than ones that are closer (Flavell, Flavell, Green, & Wilcox, 1980), objects appear bigger if they are closer (Pillow & Flavell, 1986), and the orientation of an object to an observer’s line of sight determines its projective shape (Pillow & Flavell, 1986). Absence o f Level I knowledge is basically the traditional idea of egocentrism. The research of the Flavell group has conclusively demonstrated that egocentrism in this sense simply does not exist, at least not past infancy. Children as young as 2 years will orient a picture or toy so that it can be seen by another observer (Lempers, Flavell, & Flavell, 1977). By 3 years, children understand the essential nature of hiding (Flavell, Shipstead, & Croft, 1978; Hobson, 1980; Hughes & Donaldson, 1979), can easily predict what observers see when they look at the opposite side of a barrier from the child (Flavell, Everett, Croft & Flavell, 1981; Gullo & Bersani, 1983; Masangkay et af.,1974), can infer what observers see from noting their eye position (Masangkay et al., 1974), and know what observers see in displays when what they see is specifiable in terms of discrete features, such as colored spots, or fronts and backs of objects (Masangkay et al., 1974; Verkozen, 1975). When children cover or close their eyes, they may deny that others can see them-not, however, because they do not understand the other’s visual experience, but because “see you” means seeing the face region (Flavell, Shipstead, & Croft, 1980). Level 1 rules can be summarized, perhaps, by the rule that observers see whatever can be connected by a direct line to their eyes. But when all objects in an array can be viewed by observers from a variety of positions, as is true in most perspective-taking tasks, such a rule will fail to predict a unique view. In this case, Level 2 rules become relevant, as does actual computation of other views. The developmental relationship of acquisition of Level 2 rules to acquisition of the ability to compute is not clear. One hypothesis is that Level 2 rules are prerequisite to some further development that allows successful completion of the standard perspective-taking task. Flavell, Omanson, and Latham (1978) in fact hypothesized that Level 2 rules would be prerequisite to computation. Unfortunately, their data provided little support. Performance on problems solvable only by rule was not correlated with performance on problems solvable only by computation. Importantly for the sequence hypothesis, several children who showed no understanding of Level 2 rules performed perfectly on computational problems. At what age are Level 2 rules acquired? Salatas and Flavell (1976) found considerable improvement between age 6 and 8 years in understanding of

210

Nora Newcombe

Level 2 rules. However, Flavell et al. (1978) found less evidence of developmental differences with 6-year-olds doing better than they had with the slightly different methodology used by Salatas and Flavell. Subsequent studies have converged on the view that Level 2 rules are acquired at around 4 or 5 years of age (Flavell, et al., 1980, 1981; Pillow & Flavell, 1986). This age seems too early for Level 2 rules and perspective taking to be related, if one assumes perspective taking is difficult until 9 or 10 years. However, perspective taking with item questions (Huttenlocher & Presson, 1979) is a task requiring computation, but it is much easier than the traditional task with appearance questions. Recent evidence shows that children as young as 5 years of age d o well with item questions (Newcombe, 1989). Further work will be necessary to determine the ages of first success on these tasks and to compare these data to data on Level 2 rules. on these tasks and to compare these data to data on Level 2 rules. In summary, a fair conclusion at present is that by age 2 or 3 years, children know that other observers see different objects from those that they see and that these objects can be inferred by looking across a direct line of sight from the observer’s eyes. In this sense, toddlers are clearly not egocentric. By age 4 or 5 years, children have mastered a number of additional rules regarding others’ visual experience. The relationship of these rules to the computation of the particular nature of that experience is not clear. One hypothesis-that of Piaget and Inhelder as well as others-is that the ability to compute others’ views depends on the nature of children’s systems for encoding the location of objects in space. In this view, the acquisition of Level 2 rules would bear no necessary relationship to the ability to solve problems requiring computation. C. OPERATIVE DEVELOPMENT

Piaget’s theory is a stage theory of human development with cognitive development described as progressing through four stages: sensorimotor, preoperational, concrete operational, and formal operational. Critics of the stage concept have focused on the fact that successful performance on the various tasks used to index the stages often does not appear at about the same time in development (e.g., Flavell, 1971; Gelman & Baillargeon, 1983). Perspective taking has not, by and large, figured in this debate, although successful visual perspective taking is often written about as an aspect of operative development (e.g., Brodzinsky, 1982). Hobson (1982) is one exception: He sought to assess the relationship of perspective taking to concrete operations directly by giving children aged 3 to 7 years tests such as conservation of number as well as perspective-taking tasks. He found that younger children, who failed the concrete operational tasks, nevertheless succeeded on the perspective tasks. However, the perspec-

Development of Spatial Perspective raking

21 1

tive tasks were simple Level 1 situations, not necessarily requiring use of projective space. The hypothesis that achievement of perspective-taking skills should occur at the same time as the advent of concrete operations was thus not really tested. In any case, the hypothesis that perspective taking is an aspect of the development of concrete operations seems not exactly to be something Piaget really claimed. Piaget’s thinking about the link between understanding of projective and Euclidean space and concrete operations is fairly complex. On the one hand, he considered the two entities to be fundamentally different. Spatial understanding was characterized as sublogical and different from the logical understanding that was the core of Piaget’s theory: Concrete operations of a logico-arithmetical character deal solely with similarities (classes and symmetrical relations) and differences (asymmetrical relations) or both together (numbers) between discrete objects in discontinuous wholes independent of their spatio-temporal location. Exactly parallel with these operations there exist operations of a spatio-temporal or sub-logical character, and it is precisely these which constitute the idea of space. (Piaget & Inhelder, 1948/1967, p. 450, italics in original)

On the other hand Piaget believed that sublogical and logical systems are related in several ways. The first is that at a formal level, the systems could be considered the same. [Sublogical operations] substitute the concept of proximity for that of resemblance, difference of order or position (especially the concept of displacement) for difference in general, and the concept of measurement for that of number. Once expressed in propositional form they are indistinguishable from logico-arithmetical operations, of which they constitute merely a particular species, that of continuity as opposed to discontinuity operations. (Piaget & Inhelder, 1948/1967, p. 450)

Similarly, as the two systems develop, Piaget emphasized the formal parallelisms between them. Thus, much of the final chapter of The Childs Construction of Space was concerned with the delineation of the sublogical operations of topological, projective, and Euclidean space, in terms of concepts such as addition of elements and sets and one-to-one multiplication of elements. Piaget and Inhelder did not, however, hypothesize either an exact formal relationship between concrete operations and the acquisition of projective and Euclidean space, or an exact correspondence in developmental time. Thus, for instance, the first topological grouping they discussed-partition of sets and addition of subsets-was said to be “the exact equivalent of class inclusions (A,B,C, etc.) in logic” (Piaget & Inhelder, 1948/1967, p. 463, italics added). For each of topological, projective, and Euclidean space, in fact, they delineated eight operational groupings, with similarities among them but also differences (an example of vertical decalage). They specifically noted the

212

Nora Newcombe

existence of eight concrete operational groupings, writing that “this correspondence is very interesting as regards the functional unity of the various operations of thought” (Piaget & Inhelder, 1948/1967, p. 480). Thus, the use of groupings to describe spatial thought does not appear to imply an empirical relationship between logical and sublogical development. Piaget and Inhelder did note certain empirical relationships between logical and sublogical development, however. In particular, concrete operational thinking seemed to begin somewhat earlier than understanding of projective or Euclidean space. Piaget and Inhelder hypothesized that topological relations become organized as operational systems at about the same time as operational systems emerge for logical classes and relations, or perhaps a little later because spatial relations are continuous rather than discontinuous. The structuring of topological relations, by the age of about 7, in turn allows rapid development of the projective and Euclidean systems by the age of about 9 or 10. In The Child‘s Conception of Geometry, Piaget, Inhelder and Szeminska (1948/1981) continued their investigation of spatial development, studying children’s ideas of measurement and understanding of the conservation of distance, length, area, and volume. They argued that the construction of Euclidean space is a gradual process accomplished in three stages. The first stage involves the qualitative understanding of conservation of distance, length, area, and volume, and the use of transitive reasoning when using one object to compare the size of two others. Only at the second stage does true measurement emerge, but even then, area and volume cannot be calculated. The third stage is not attained until formal operations. In summary, Piaget’s hypothesis of a concrete operational stage has different meanings when applied to content involving classes and relations than when it is applied to spatial content. Spatial content can be topological, projective, or Euclidean; topological content can be structured either intuitively or operationally. Although some aspects of logical development are linked to some aspects of spatial development (e.g., transitivity with a qualitative understanding of measurement), in general, the links between logical and spatial development are more formal than empirical.

111. How Can Spatial Location Be Encoded? A. TOPOLOGICAL, PROJECTIVE, AND EUCLIDEAN SPACE

%pO/OgkQlspace is defined by Piaget and Inhelder as encoding relationships of enclosure, touching, order, proximity, and separation within objects. Another way of defining topological space is that it encodes spatial relationships that would be maintained under elastic distortions. It is thus nonmetric.

Development of Spatial &rspective Taking

213

It fails to distinguish between straight lines and curved ones, and between curves and angles. Projective space differs from topological space in several ways. First, it does encode relationships among separate objects. Objects are related in terms of straight lines “projected” from one to the other. One of these objects may be the self, in which case objects are linked to the viewpoint of the subject, or one of the objects may be other viewers. Thus, projective space entails the realization that viewers in different positions have different views, as well as the ability to identify the view of each. These abilities are referred to together as the ability to coordinate views or perspectives. The idea of a straight line is itself a nontopological one. In topological space, which considers only order, straight and curved lines are equivalent. Piaget and Inhelder suggested that the fact that different interpositions of objects are seen when arrays are viewed from different positions is part of how children form the idea of a straight line. Thus, the understanding of points of view and the idea of a straight line are thoroughly interdependent notions. In summary, locations are coded in projective space as falling or not falling on straight lines drawn through specific landmarks or viewing points. Euclidean space refers to the encoding of spatial location using metric coordinates. Locations are represented relative to an abstract system, rather than relative to each other as in projective space. An ambiguity in Piaget and Inhelder’s writing is whether projective and Euclidean space are “separate but equal,” or whether projective space is simpler than Euclidean space and a developmental prerequisite for it. On the one hand, Part Three of their book is entitled “The Transition from Projective to Euclidean Space,” and the synopsis includes a discussion of transitional stages between projective and Euclidean space. Later, Piaget and Inhelder noted: We have shown that projective concepts imply a comprehensive linking together of figures in a single system, based on the coordination of a number of different viewpoints. But side by side with the development of this organized complex of viewpoints there also takes place a coordination of objects as such. This leads ultimateely to the idea of Euclidean space, the concepts of parallels, angles and proportion providing the transition between the two systems. (p. 375, italics added)

On the other hand, projective and Euclidean relations were described as being constructed “concurrently” and “side by side,” and as being “mutually interdependent” (all phrases on p. 419). Piaget and Inhelder later stated: [Concurrent development] is hardly surprising since, on the one hand, a system of reference embraces a view of the whole from a particular viewpoint (remote enough for perspective lines to be parallel). On the other hand, individual perspectives (45”, for instance) are relative to the positions of the observer and the objects, and this the subject can only express by locating them within a total spatial field

Nora Newcornbe

214

defined by a coordinate system. Whether the first is achieved before the second or the other way round, sooner or later one will inevitably react upon the other. (P. 441)

One might be tempted to conclude that Piaget and Inhelder really meant one or the other of these two apparently contradictory positions. Apparently, however, their accounts of this point involve an unresolved tension. This tension revolves around their preference for Euclidean space as an endpoint of development, in that the achievement of an abstract coordinate system represents the attainment of an understanding of space itself as an absolute entity independent of whether or not it is filled by objects (the “container view” of space). However, even an abstract system, to be useful, must, as they themselves recognized, be anchored in space through the use of certain locations as defining axes or coordinates. Euclidean space uses “certain favored positions as reference points or “points of departure,” for all subsequent positions” (p. 376). Thus, the two systems, one absolute and the other relative, are in practice quite similar, and developmentally Piaget and Inhelder believed they are acquired together. B. CRITICISMS

I.

Terminological

Tho terminological points need to be kept in mind in reflecting on Piaget and Inhelder’s proposals about spatial representation. The first is that their usage of topological, as pointed out by Mandler (1983), differs from that of mathematicians. The second is that the term Euclidean is also infelicitous in some regards, in that Euclidean geometry is not concerned with metrics. Possibly “Cartesian space” would have been a better choice, embodying the idea of coding location in terms of a coordinate system. However, if the definitions given by Piaget and Inhelder are kept in mind, the choice of terminology should not get in the way of a serious consideration of the merits of their theory.

2. Logical The definition of topological space used by Piaget and Inhelder contains a very important problem-whether proximity can really be considered a topological notion. Piaget and Inhelder included proximity on their list of topological concepts (e.g., p. 153). But points being “near” to one another seems to involve some appreciation of metrics, and topological space is supposed to be fundamentally nonmetric. One way out of this conundrum might be that points are considered proximate if they form units according to the Gestalt laws of grouping. Because topological proximity is defined only within objects, not between them,

Development of Spatial Rrspective Taking

215

principles such as common fate would not apply but principles such as similarity might be relevant. This hypothesis would entail predicting that the similarity of elements in a pattern or parts of an object would affect whether or not they were seen as neighbors by young children. Of course, an alternative is that young children actually are capable of encoding distance in at least a rough way, and that thus proximity does have a metric meaning quite early. We know surprisingly little about the encoding of distance in infancy and early childhood so the question does not have a current empirical answer. (See Newcombe, 1988, for further discussion of the problem of proximity.)

3. Empirical Topologicalspace, as Piaget and Inhelder defined it, does not distinguish among enclosed areas, so that angular shapes such as squares and triangles are equivalent to circles or ellipses. But as Piaget and Inhelder themselves found, children do make such distinctions, at least by the age of 4 years. Distinctions between curvilinear and rectilinear shapes were also found by Love11 (1959), Laurendeau and Pinard (1970) and Rieser and Edwards (1979). A second problem with the hypothesis that young children encode space only topologically is that topological space was defined by Piaget and Inhelder as encoding relationships only within objects, but not relationships between them. A large literature, however, indicates that even infants can use between object coding, as when, for example, 9-month-olds use the mother as a landmark for an interesting event occurring to one side of her (Presson & Ihrig, 1982). The more common interpretation of Piaget and Inhelder has therefore been that topological coding does occur between objects, but that the absence of a metric coordinate system leads to great reliance on the presence of landmarks for encoding spatial location. Thus, for instance, Acredolo, Pick, and Olsen (1975) found that younger but not older children remembered the location of an object more accurately with a landmark present, and Herman and Siege1 (1978) found that kindergartners but not older children remembered the layout of a model village more accurately with landmarks close by rather than distant. Huttenlocher and Newcombe (1984, Experiment 1) found, however, that younger children benefited more than older ones from the provision of landmarks only when metric accuracy was considered. Even 5-year-olds in nolandmark conditions showed excellent recall for the relationship among a set of objects (i.e., had high configurational accuracy). Thus, young children were able to code spatial location even in the absence of nearby landmarks, but they did so in a rather inexact fashion. At least by 5 years, then, even a relatively weak prediction of the topological space hypothesis is not confirmed. Somerville and Bryant (1985) and Bryant and Somerville (1986) have shown

216

Nora Newcombe

that 6-year-old children can readily use coordinate systems for encoding location in paper-and-pencil tasks, and that even 4-year-olds’ performance is significantly above chance. Their studies were not addressed to the ability to impose an abstract spatial coordinate system, especially in a large-scale space, but they do indicate that the ability to exploit a metric coordinate system is present considerably earlier than Piaget and Inhelder had believed. Perhaps the most radical questions about the primacy of early topological coding have been raised by Landau (1988; Landau, Gleitman, & Spelke, 1981). The ability of a 2-year-old blind child to traverse novel routes in a simple environment was argued to indicate the innateness of Cartesian space. Methodological questions about this work, however, preclude concluding that children this young use Euclidean space, let alone that such spatial concepts are innate (Liben, 1988). In summary, several empirical grounds lead to doubt that children, at least by the age of 4 or 5 years, are incapable of coding space in any way other than topological. They appreciate distinctions not recognized in topological space; they code location fairly well even in the absence of nearby landmarks; and they can exploit a coordinate system provided by an experimenter. However, the most radical alternative, that Cartesian space is innate, lacks sufficient evidence to be accepted. Alternative conceptualizations of how location is encoded and how this encoding might change developmentally seem to be needed. C. ANOTHER MODEL OF LQCATION CODING

Huttenlocher and Newcombe (1984) have proposed a model of the development of location coding with several major steps. At first, the child remembers the location of small target objects only when they are coincident with larger fixed landmarks. (This process is more like paired-associate learning than it is like location coding.) Later, by about 2 years, children remember the location of targets in terms of a rough estimate of their distance from a landmark. Such a coding is likely to be in terms of proximity or neighborhood, a concept which, as noted, is included in Piaget and Inhelder’s definition of topology, but which seems fundamentally metric (Newcombe, 1988). A third step is taken by at least 5 years of age, and perhaps earlier, when children appear to use more than one landmark (i.e., frameworks of landmarks) to encode location. This is a much more exact system than using single landmarks. If the framework is constant for all targets encoded (common framework), a framework of landmarks approximates a coordinate system, and an empirical distinction between the two systems may be difficult to draw. Whether or not common frameworks are used as soon as individual frameworks are used is an empirical question. A logical possibility is that the child at first uses different frameworks for different targets and only later

Development of Spatial Perspective Eking

217

realizes the virtues of using a common framework. Certainly, a common framework is apparently used by age 8, given children’s ability at that age to answer spatial transformation questions requiring such coding (Huttenlocher & Presson, 1979). Recent evidence indicates that common frameworks are also present by age 5 , again given the ability at that age to answer spatial transformation questions (Newcombe, 1989). A second issue in spatial coding concerns what Huttenlocher and Newcombe termed “the internal coding of an array of targets”-that is, remembering the relationships among the targets so that the whole is seen as forming a specified shape. In a sense, internal coding involves transforming a set of separate objects into a single object with vertices. Such a coding might make spatial transformation tasks easier, because it might allow all the targets to be transformed as a unit, in a mental rotation process (e.g., Shepard & Metzler, 1971). However, patterns of difficulty on various spatial transformation tasks indicate that neither 8-year-olds nor adults form or use such encodings (Huttenlocher & Presson, 1979; Presson, 1982). Two-year-old children appear to have particular difficulty with internal coding, being unable to utilize it even when it is provided for them by the experimenter (Newcombe, Dubas, & Spies, 1985). In summary, coding of single targets may develop from association with single landmarks, to proximity to single landmarks, to distance from frameworks of landmarks, with the third stage possibly including a sequence from local to overall frameworks. Internal coding develops more slowly, and is often not used even by adults and even in situations where it would be helpful.

IV. Factors Affecting Success on Perspective-Taking Tasks This section contains a review of the literature on which factors make perspective taking harder or easier. It is organized into four major sections, dealing with attributes of the subject, the task, the display, and the response mode or dependent variable. The section has two overall aims: first, to summarize what is known about the perspective-taking task in general; and second, to evaluate what variations in performance reveal about spatial representation in particular. A. ATTRIBUTES OF SUBJECT

1. Intellectual Realism

Piaget and Inhelder’s original version of the three-mountains task showed children the mountains so that all were visible from the child’s viewpoint, but

218

Nora Newcombe

some mountains were hidden from the viewpoint of others. One reason for selecting or building an egocentric view when the view of another is requested might thus be another attribute of children’s thought identified by Piaget, intelfectual realism: Children may wish to show all the objects they know to be present as present. (However, see Estes, Wellman, and Woolley, Chapter 2, this volume, for some criticism of the concept of intellectual realism.) The importance of intellectual realism in the perspective-taking performance of children 6 years and younger has been clearly demonstrated (Liben, 1978; Liben & Belknap, 1981; Light & Nix, 1983). For instance, Liben and Belknap found that 3-, 4-, and 5-year-olds had marked difficulty selecting their own view when that view showed a large block, behind which children had seen other blocks placed. Thus, investigations of perspective taking using objects that can occlude each other may confound the waning of intellectual realism with the growth of other components of perspective-taking ability. Pillow and Flavell(l985) showed, however, that the problem of intellectual realism may be confined to studies using picture selection as a response made (see Section IV,D) or studies using the expression “looks like” in verbal questioning. In terms of an interest in spatial representation, an important point is that developmental differences are still found on tasks where only a single object is used (Jacobsen & Waters, 1985) or where the display consists of small objects viewed from above (Huttenlocher & Presson, 1979; Newcombe, 1989), so that no occlusion is involved in perspective taking. Thus, the issue of intellectual realism is in one sense orthogonal to the issue of spatial representation. However, intellectual realism is an important factor to consider in methodology: If one wishes to examine the demand perspective taking makes on representation, at least with younger children, one should use displays that do not involve occlusion. In addition, one must develop a procedure in which children can correctly identify their own view, a prerequisite not always met even when displays do not involve occlusion (Gzesh & Surber, 1985).

2. Cognitive Style

n o dimensions of cognitive style have been investigated for relationships to perspective-taking ability: field independence and reflectivity-impulsivity. Field independence was found to be associated with better performance on spatial perspective taking by Bowd (1975), Okonji and Olagbaiye (1975), and Finley, Solla, and Cowan (1977), although no association was found by Knudson and Kagan (1977). Reflectivity was found to be associated with better spatial perspective taking at ages 6 and 8 years by Brodzinsky (1980, 1982). In addition, Brodzinsky (1982) found a longitudinal relationship between reflectivity at age 6 and perspective taking at age 8; structural modeling suggested this was consistent with a causal relationship.

Development of Spatial Perspective Taking

219

Two issues need to be considered with regard to these findings. The first is that the tests measuring field independence (Children’s Embedded Figures Test) and reflectivity (Matching Familiar Figures) involve spatial analysis. Thus, the relationship between these variables and spatial perspective taking may reflect that some element of spatial analysis is involved in both tasks, rather than showing that cognitive style as a construct has an effect on the development of perspective taking. Second, even if individual differences in cognitive style as such have an effect on the speed of acquisition of perspectivetaking skills, what component of the task is affected is not known. The spatial representation that underlies the ability to solve transformation problems might or might not remain constant. Thus the relevance of cognitive style to changes in spatial representation is not known. B. ATTRIBUTES OF TASK

1. Naturalism of Task

Piaget has often been criticized for using tasks and procedures unnatural to children and therefore underestimating their skills (e.g., Donaldson, 1978). With respect to spatial perspective taking, several specific factors have been mentioned. First, Piaget and Inhelder used dolls to indicate the position of the “other,” but dolls are inanimate hypothetical observers that do not really have a view at all. Consistent with this criticism, Cox (1975, 1977a) and Fehr (1979) have found better performance on perspective taking when real people rather than dolls are in the position of the other. Fehr (1979) found that performance with blindfolded people was no better than that with dolls, indicating that the advantage of using people was due to their status as true potential observers of the scene. Thus, using real people as observers facilitates performance. However, it does not eliminate errors or developmental differences, and it seems to be a variable improving performance in general rather than one revealing about developmental change. A second criticism is that the perspective-taking situation is not sufficiently like a game to motivate children, or not sufficiently familiar for them to understand what their task is supposed to be. Both Hughes and Donaldson (1979) and Hobson (1980) found excellent performance among preschool children with hide-and-seek games. A related criticism is that the usual instructions are not clear enough for young children. Hughes (1978) demonstrated that asking preliminary questions directing preschoolers’ attention to critical aspects of a display showing three dolls improved their performance considerably. Clearly, however, the tasks used in these demonstrations could be solved using Level I skills, and they did not require spatial transformation. Thus,

220

Nora Newcombe

the studies add to the forgoing large body of evidence that simple rules regarding what other people can see are mastered early, and that young children are not truly egocentric. They do not, however show that children could succeed on a task with demands comparable to the three-mountains one, if only it were embedded in a more familiar or engaging context. A third criticism of Piaget and Inhelder’s procedure has been that by asking children to demonstrate their own view first, experimenters increase egocentrism by suggesting to children who are perplexed by the problem that their own view is an acceptable answer. Indeed Aebli (1967) and Garner and Plant (1972) have shown that children not asked first about their own view make fewer egocentric errors. Interestingly, however, Aebli did not find a decreased number of total errors (and Garner and Plant did not test this). Thus this aspect of the procedure increases the proportion of errors that are egocentric, and the fact that the proportion can so easily be diminished suggests that the answer is, as Aebli put it in his title a “substitute solution for an insoluble task.” But why the task is so difficult remains unexplained. A fourth class of variables investigated by those interested in the real-world context of these problems involves whether children’s performance is better when they are first allowed either to play with the materials or to interact with them together with other children of either the same or different cognitive levels. Matthews, Beebe, and Bopp (1980) found a small facilitative effect of play with materials on perspective taking, although it may not have been statistically significant. Doise, Mugny, and Perret-Clermont (1975) and Mugny and Doise (1978) found better performance on an allocentric placement task (see Section V,A) following experience working with a partner on the problem, but this finding was not replicated by Bearison, Magzamen, and Filardo (1986) for perspective taking. Bearison et al. did, however, find better performance at least among boys who experienced a moderate amount of sociocognitive conflict in the dyads. This research at least suggests that play, either solitary or interactive, may be helpful to cognitive growth. However, even if true, we would need to specify what was being acquired. In particular, the research does not deal with whether or not play and social interaction lead to the development of more sophisticated systems of spatial coding or transformation or to more efficient or consistent use of existing systems. In summary, increasing the naturalism of the task does, in some cases, lead to better performance, or at least to reduced egocentrism. However, the research in this tradition has not been focused on process, and some experiments have involved Level 1 tasks rather than tasks requiring computation. Thus, how the naturalism of the task affects spatial coding has not been assessed and the studies do not necessarily show that younger children can encode and transform spatial location in the same way as older children, given natural tasks.

Development of Spatial Brspective Taking

221

2. Shielding of Display Several researchers have tested the hypothesis that shielding the display from view when children are asked to select the other’s view would enhance performance, presumably by reducing distracting perceptual cues. Brodzinsky, Jackson, and Overton (1972) found improved performance, at least on multiple object arrays, for 8- and 10-year-olds, but not 6-year-olds. They suggested that shielding did not help the younger children because they had no underlying understanding of the problem that could be activated in the right circumstance. Walker and Gollin (1977) found that shielding reduced the incidence of egocentric errors for 4-year-olds but not 7-year-olds on a single-object task. Shielding did not reduce total errors, however; it simply changed the type of error. These two studies suggest that shielding the array may improve performance at least for certain arrays and ages, using developmentally appropriate dependent variables. Age trends remain, however, within the shielded conditions, indicating that perceptual distraction is but one source of age differences on this task. The evidence, in any case, is not uniformly positive. De Lisi, Locker, and Youniss (1976) failed to find an effect of shielding, also using multiple-object arrays and the same age groups as Brodzinsky et al. Flavell, Botkin, Fry, Wright, and Jarvis (1968) did not find that having children turn away from the array before answering improved their performance, although this might be because turning away not only removes the array from view but also introduces a distracting rotational component to the problem. Most problematic is that Huttenlocher and Presson (1973) actually found the opposite effect: Shielding the array led to worse performance. They suggest memory demands were increased when the array was not visible. In summary, the effect of shielding is most likely a complex one. It may be helpful to certain age groups, but probably only when the tradeoff in terms of extra memory demands is not too severe.

3. Experience with Array from Different Viewpoints Eiser (1974) noted that some investigators allow children to walk around the display before being asked questions, in which case the questions may tap children’s memory for the various views, as well as their ability to infer the views. Her own study of 7-year-olds showed that walking once around the array led to a slightly smaller number of egocentric errors than shown by an inference group but no significant reduction in overall errors. More experience might of course lead to more marked effects. But one would have to be careful not to design a task that assesses memory, which would not be relevant for understanding spatial transformation.

222

Nora Newcombe

4. Actual versus Imagined Movement The classic perspective-taking task involves spatial inference, with the subject remaining in one position but attempting to construct the view from another. Several researchers have asked what performance would be like if subjects were asked to move to another position, of course with the array covered so that the answer would not be self-evident. The evidence seems quite clearly to indicate that physical movement makes the task much easier. Shantz and Watson (1971) found excellent performance by 4-, 5-, and 6-year olds on such a task; the same subjects, however, had great difficulty on classic perspective taking. The same results were obtained by Huttenlocher and Presson (1973), testing 9-year-olds, and Schatzow, Kahane, and Youniss (1980), testing 8- and 10-year-olds. More recently, Huttenlocher and Newcombe (1984, Experiment 2) showed that even 2-year-olds could locate an array of objects correctly after moving to the opposite side of a board, as long as spots marked the correct vertices of the array. Why are objects so easy to locate correctly after actual physical movement, and so difficult when the movement must be imagined? Huttenlocher and Newcombe (1984) suggested that the reason may be that locations of objects are typically coded in relation to a framework of external landmarks. When the subjects move, they change their relation to that framework, so that the location of the targets becomes perceptually obvious and minimal inference is involved.

5. Availability of Outside Landmarks Given the argument in the preceding section, perspective taking should be better when external landmarks are available than when they are not. Unfortunately, no existing study has tested exactly this hypothesis. Fehr (1980), Fehr and Fishbein (1976), Huttenlocher and Presson (1973), and Lapsley, Fehr, and Enright (1981) all found better performance when an additional landmark was added, presumably with background landmarks in the testing rooms always present. The nature of the added landmark is not clear in the latter three reports; Huttenlocher and Presson used a horse in a fixed position facing the array. P. L. Harris and Bassett (1976, Experiment 1) failed to find an effect of including a small shell as a landmark on the child’s board in a modelbuilding task. Presson (1980) conducted the most direct test of the hypothesis that locations in the target array are usually coded with respect to the wider spatial field. He compared perspective-taking performance under standard conditions and performance under a modified condition in which children observed a model of the experimental room, oriented as it would be seen by the hypothetical viewer. This cue improved the performance of 7-, 9-, and ll-year-

Development of Spatial Pmpective Taking

223

olds. It also decreased egocentric errors for 90"and 270" movement, although not for 180" movements. Thus, the evidence clearly indicates that through middle childhood, the spatial locations of subjects in perspective-taking arrays are coded with respect to external landmarks.

6. Amount of Movement Investigators have generally compared the difficulty of imagining an observer on the opposite side of the array (180" movement) with the difficulty of 90" and 270" movement, although some work has included other positions as well. Some researchers have speculated that 180" movement might be easier to imagine than other amounts, for one of several reasons. When fronted objects are used in the array, and shown facing the child in the 0" position, recognizing the fronts and backs of ths objects may be easier than recognizing the sides (Gzesh & Surber, 1985). In addition, when objects are shown so that movement results in varying degrees of occlusion and interposition, recognizing the reversal of what is visible or hidden at 180" may be easier than recognizing the differences as seen from the side views (although this argument is at odds with the intellectual realism argument). Finally, the reversal of dimensions at 180" (front-back and left-right reversal) might be easier to compute than the cross-dimension changes occurring at side views (left becomes front, front becomes right, etc.). Other investigators have hypothesized that 180" movement should be harder than all other conditions, for one of two reasons. If one believes that subjects reposition themselves or locate the other observer by continuously tracking a path from their current position in their "mind's eye" and that they take the shortest route to do so, going either left or right, depending on the distance to the target location, then 180" represents the greatest distance traveled, and errors might increase with distance. Another idea focuses on the fact that dimensions remain constant at 180", but suggests that reversal is confusing for subjects rather than facilitating. Some studies have shown 180" movement to be easier than other amounts of movement (Eiser, 1974; Gzesh & Surber, 1985; Presson, 1982, Experiment 2; Rosser, Ensing, Mazzeo, & Horan 1985, rotation task), but other researchers have found significantly greater errors at 180" (Cox, 1977b; Fehr, McMahon, & Fehr, 1982; Nigl & Fishbein, 1974). Jacobsen and Waters (1985) found a trend for more errors at 180" for a task involving the position of a single object. The conservative conclusion to this controversy is that difficulty does not systematically vary as a function of observer position. The number of studies showing effects each way is roughly the same. Other researchers found no significant differences (Bialystock, 1986; Borke, 1975; Cox & Willetts, 1982;

Nora Newcombe

224

Presson, 1980, 1982, Experiments 1 & 3; Rosser et al., 1985, construction task; Schachter & Gollin, 1979; Shantz & Watson, 1971). Many other researchers did not report data separately by observer position, and at least in some cases, this omission may be because preliminary analyses showed no differences. From a theoretical point of view, the reasons given for the putative ease or difficulty of 180" movement seem contradictory and post hoc.

7.

Training

Most Piagetian tasks have been the subject of extensive efforts to train better performance in young children, thereby obtaining clues about what experiences they lack or what hidden competencies they have. Perspective taking, for no apparent reason, has been an exception, but some attempts have been reported. Miller, Boismier, and Hooks (1969) found a modest improvement in perspective-taking performance in 7-year-olds, following six individual lessons on sighting, transformations along left-right and near-far dimensions, and practice on perspective tasks of graded difficulty. Cox (1977~)found that 5-year-olds showed delayed as well as immediate improvement on perspective and related tasks, following 20 individual training sessions on tasks of gradually increasing difficulty, with verbal feedback from the experimenter and movement by the child to check responses. An especially notable feature of Cox's study was the very high performance levels reached and maintained by the young children. Both Cox and Miller et al. developed training programs dealing with as many components of perspective taking as possible. Silverman (1986) found significant improvement in perspective taking for 6 , 8-, and 10-year-old children following simple feedback on the correctness of answers, including better performance on delayed tests and transfer tests. This finding suggests that at least part of children's difficulty may be simple lack of understanding of the procedure. That this problem is not the whole story, however, is shown by the fact that performance, especially on transfer tests, was still far from perfect, as well as the fact that age differences persisted within conditions. Perhaps the most interesting aspect of Silverman's study from the point of view of spatial representation lies in the fact that instruction in coding the array failed to improve performance, and in fact, when feedback was absent, actually seemed to hurt performance. The instructions emphasized coding the array as a unit, something subjects apparently rarely do and may find difficult (Huttenlocher & Presson, 1979; Presson, 1982). Kielgast (1971) also found that having children verbally describe the internal relationships of array items was not helpful to, and not correlated with, their performance on perspective taking. It would be interesting to see if instructions emphasizing locating objects in relation to the external framework would improve

Development of Spatial Perspective Taking

225

performance on tasks such as item questions for which this coding is an asset (Huttenlocher & Presson, 1979). In summary, perspective taking appears to improve following training, but training studies to date have not been analytic regarding the components of the task being taught. We need to know what specific skills or coding schemes children most need instruction in, and whether these apparent lacks are the same or different at different ages. C. ATTRIBUTES OF DISPLAY

1. Characteristics of Objects

The objects composing the visual display shown in perspective-taking tasks may be familiar or unfamiliar to the child, may be symmetric or have canonical fronts, may vary or not vary among themselves in perceptual characteristics such as height or color, and so on. In addition, the objects may be arranged in symmetric patterns with respect to each other (squares, triangles, circles) or be arranged asymmetrically; if the objects are meaningful and familiar to children, arrangements may be themselves meaningful (e.g., barnyard scene) or not. Surprisingly little parametric research exists on the effect of such manipulations on perspective-taking ability despite the frequency of speculation on the subject. Thus, for example, the difficulty even 11- or 12-year-old children had with some of Laurendeau and Pinard’s (1970) tasks is often attributed to those researchers’ use of symmetric, unmarked objects in displays, but no systematic comparisons support this commonsense argument.

a. Familiarity Familiarity has not been investigated in a fashion unconfounded with symmetry. Borke (1975) found that when the display involved familiar fronted objects such as animals and houses, 3- and 4-year-olds had less difficulty indicating the other’s perspective with a model rotation device than with Piaget and Inhelder’s mountain display. This finding was obtained even when the number of familiar objects was much greater than three. However, the familiar objects were fronted, and the mountains were not (except that a small house appeared on one mountain). A similar comparison made by Rosser et al. (1985) yielded different results. They used either animals or vehicles (objects that are both familiar and fronted) for their “marked” displays and rectangular solids and circles for their “unmarked” ones. (Number of colors and arrangement of objects were also confounded with “marking” in this study.) Rosser et al. found no difference between the two display types when children used a model rotation device, in contradiction to Borke. (In addition, when the task was model construction, the familiar fronted objects were less accurately placed than the

226

Nora Newcombe

geometric solids, but this comparison is flawed by the fact that children’s placements were scored by different criteria in the two conditions: marked objects had to be oriented correctly as well as to occupy the correct position to receive a point.) Thus, although familiar objects may well be easier to encode and transform, for reasons having to do with motivation or perceptual differentiation or both, comparisons are needed that unconfound familiarity with symmetry before this point can be accepted.

b. Orientation. The work of Borke and Rosser et al. just discussed deals with frontedness of objects as much as with familiarity. One might assume that symmetric objects would be harder to deal with in perspective taking then fronted ones because symmetric objects might be confusable with each other. However, reflection suggests that the influence of symmetry depends on the response required. If the correct choice from a given point of view must be selected from a set in which all choices preserve the internal structure of the display, then having fronted objects may well be advantageous. Given fronted objects, a child may be able to choose correctly by attending to one specific feature-figuring out, for instance whether the horse would be seen head on, back on, or side on by another observer. In this case, of course, the child is not really required to consider the spatial location of the objects or to transform the locations, and thus this task does not assess issues of location coding at all. If the child must consider orientation as well as position in making a response, however, as is required when reconstructing a scene from another’s perspective, then the use of fronted objects may make the task more difficult. Encoding the orientation of objects has been studied as a topic in its own right. L. J. Harris and Strommen (1972) found a high degree of consensus among 4- to 7-year-olds in their interpretation of directions involving “front,” “back,” and “beside.” This was greater for featured objects, but still considerable for featureless (symmetric) objects. Subsequent research has shown that children as young as 2 years have acquired the concepts of front and back, even when they have not mastered the linguistic terms (Levine & Carey, 1982). Thus, children can encode orientation early, but less is known about the development of the ability to mentally transform orientation. Encoding the orientation of objects in perspective-taking tasks has been studied by Coie, Costanzo, and Farnill (1973) and Eiser (1976). Coie et af. conducted analyses of the kinds of errors made by children from ages 5 to 11 years on picture choice tasks. They concluded that children are able to transform orientation (which they called “aspect”) by age 10, but that children of this age still have difficulty with position on the right-left dimension. Thus,

Development of Spatial Perspective Eking

221

by implication, they supported the foregoing argument that featured objects would enhance performance on perspective-taking tasks in which figuring out the correct orientation of at least one object allows correct choice, but that such choice does not necessarily assess the ability to transform spatial location and to infer position. The data of Eiser (1976) also suggest that orientation and position of fronted objects may be attended to differentially, although they initially seem to undercut the argument that orientation information can be used to enhance performance on perspective-taking tasks by allowing position to be ignored. Children of 8, 10, and 12 years had to guess what an experimenter hidden from them was seeing by asking her questions. Questions were coded as regarding either the orientation of one of the three objects (two were fronted) or as regarding their position. Success at guessing the perspective of the experimenter was greater when position questions were asked, and less with orientation questions. Of course, children themselves selected their questions, so less cognitively advanced children may have concentrated on orientation questions and more advanced children on position questions. It would be interesting to know how children would respond in a guessing paradigm where hints of each kind were supplied by an experimenter. In this case, orientation information might be more helpful than position information, or helpful at younger ages.

c. Other Characteristics. Fehr, Lapsley, Enright, McMahon, and Ackerman (1983) noted that perspective-taking tasks often require that children recognize three-dimensional displays from two-dimensional pictures. The mapping from three to two-dimensional stimuli might be a source of difficulty. Their data indicated that, indeed, when two-dimensional arrays were presented, subjects from 7 years of age through college showed better performance than they did with three-dimensional arrays. Children of 5 years were at low levels of performance overall. Strong age trends were present, however, within each condition, with little indication of age-related interactions, except for the floor effects with 5-year-olds. Thus dimensionality crossing does not seem to be a major confound in developmental investigations of perspective taking. Similarly, Eliot and Dayton (1976) found that board shape, block arrangement, and block shape had effects, some of them in complex interactive fashion, on the difficulty of perspective taking. These effects were apparent for subjects from 6 years of age through college, with no trace of age-related interactions. Thus developmental investigators are free to choose board shapes and block arrangements without fearing that the difficulty level set by these decisions will be different for different ages.

228

Nora Newcombe

2. Number of Objects At first thought, increasing the number of objects in the array seems likely to increase the difficulty of the task; more locations would have to be encoded and transformed. For construction tasks, increasing the number of objects has this effect. Rosser et al. (1985) have confirmed that children aged 6 and 8 years have more difficulty reconstructing four, as compared to two-object arrays. For picture-selection tasks, however, number of objects should not, logically, affect performance, because working out the position of any one item allows for correct choice unless the choices include scramblings of the internal arrangement. Similarly, performance of model-rotation tasks should not depend on number of objects in the array. Fishbein, Lewis, and Keiffer (1972) found, however, that performance was better for one object than for three objects on both a picture-selection task (Experiments 1 and 2) and a modelrotation task when ceiling effects were avoided (Experiment 2). The advantage of single objects did not vary in size with age across the range of 3 to 9 years; thus, even older children did not realize that the three-object condition was logically reducible to the single-object condition. Gzesh and Surber (1985) also found better performance for one than for three objects in a picture-selection task. Several other studies, however, have failed to yield an effect of number of objects in the array on the accuracy of picture selection (Brodzinsky et al., 1972; Liben, 1978; Minnigerode & Carey, 1974; Nigl & Fishbein, 1974). Brodzinsky et al. did, though, apparently find an advantage for single-object over multiple-object arrays in their unshielded condition. One reason that has been suggested for some of the failures to find an effect of number of objects is a threshold effect. Neither Liben nor Nigl and Fishbein included a single-object condition, and Minnigerode and Carey studied the effect of having one, two, or three landmarks on the slopes of an everpresent mountain. Once the number of objects in an array is two or more, Nigl and Fishbein argued that children may attempt to encode internal spatial relationships. The addition of a third object then provides only redundant information, and does not increase the number of spatial relationships to be encoded. This explanation is puzzling, however, in that it attributes to children a strategic appreciation of redundancy, exactly what would seem to be lacking in the failure to concentrate on the location of a single item in all cases. The idea that single-object arrays are relatively easy is challenged by the findings of Jacobsen and Waters (1985). They used a symmetrical tower as a stimulus, and found 7-year-olds having error rates as high as 68%. This contrasts with 41% errors made by 6-year-olds and 26% by 8-year-olds in the most analogous single-object condition in the Fishbein eta/. study. The

Development of Spatial Perspective raking

229

difference between the studies raises the possibility that the results of Fishbein et af. and Gzesh and Surber, and the partial replication by Brodzinsky et af. depend on the use of fronted objects. The easiness of their single-object conditions seems likely to be predicated on the fact that success can be achieved by noticing whether another observer would see the front, or back, or a particular side feature of a differentiated object. This does not require the encoding or transformation of information about spatial position. This hypothesis still leaves as a mystery the question of why children as old as 9 years are nonstrategic in dealing with orientation information; they do not simply work out the orientation of one object, but appear to try to deal with several or all of the objects in multiple-object arrays. When position information must be processed, as when objects are symmetric, increasing the number of objects may not increase difficulty. No such effect was found in studies by Liben, by Minnigerode and Carey, and by Nigl and Fishbein; Jacobsen and Waters found that a single-object task was quite difficult, even though they did not specifically compare performance to that in a multiple-object condition. Thus, in position tasks, children may in fact concentrate on the position of a single object, with the addition of other objects not increasing the difficulty.

3. Near-Far versus Left-Right Piaget and Inhelder noted that children in transitional stages seemed to be able to work out what positions would be near to or far from another observer before they could work out what positions would be to the left or right. They attributed this difference to the fact that “intuitively-which is to say, egocentrically-there is a bigger difference between a background beyond the reach of immediate action and a foreground directly subject to it, than there is between a left and right which are equally near or distant” (p. 235). Several subsequent studies have confirmed the greater difficulty of leftright relationships as compared to before-behind ones (Coie et al., 1973; Cox, 1978a, 1978b; Liben, 1978; Nigl & Fishbein, 1974). This difference is present when only one dimension must be considered at a time, as well as when both must be considered simultaneously (Hoy, 1974; Minnigerode & Carey, 1974). Furthermore the difference is not reducible to general left-right confusions (Liben, 1978), which, in any case, may have been overestimated (e.g., Braine & Eder, 1983). All the studies confirming the greater difficulty of computing left-right relationships, however, involved arrays that, like the three mountains, contain interposition. In this case, correct choices on the before-behind dimension

230

Nora Newcombe

involve knowing what will be hidden and what will be visible from a given position, rather than simply being able to work out relative spatial position. When this factor is eliminated, the difference apparently disappears. Cox and Willetts (1982) eliminated interposition by using flat arrays, and Jacobsen and Waters (1985) by using a single object. Neither study indicated a special difficulty with left and right. In summary, children initially have a preference for showing all objects as visible, but by age 7 or so they are able to predict accurately what objects will be visible from a specified position and what objects will be hidden. This ability is not, however, due to the practical difference between being able to reach an object or not (near vs. far), as shown particularly clearly by Jacobsen and Waters. Children may have difficulty on certain tasks with both leftright and near-far positions, when interposition is avoided, at least until the age of 9 or 10 years. D. RESPONSE MODE

Piaget and Inhelder reported data on three tasks: model building, picture selection, and placing of a doll to indicate the position from which a specified view could be seen. Their discussion emphasized the similarities in findings using the three tasks. Subsequent researchers have largely ignored the dollplacement task, although Miller (1967) and Laurendeau and Pinard (1970) have confirmed that the approximate ages at which it is mastered correspond to those for picture selection, with doll placement possibly somewhat easier. Investigators have, however, added several new response paradigms to the original three, as well as exploring the effect otvariations in the type and number of choices in the picture selection task.

1. Surprise Paradigms Shantz and Watson (1970, 1971) showed that children as young as 3 years gave verbal and facial indications of surprise when, after moving to another position around a covered array, they were shown the same array they had originally seen rather than a different one. This surprise shows, of course, that they expected to see something different from their first view, and thus that they are not egocentric; they know that different observers have different views. This research fits well with research reviewed earlier showing that young children are not fundamentally egocentric. Further, it suggests that, at least when children actually move to a new location, they understand a Level 2 rule, in Flavell’s terminology (i.e., different positions, different views) by the age of 3. However, the surprise paradigm does not allow for any assessment of what children expected to see in particular, and so it does not bear on issues of spatial representation.

Development of Spatial Brspective Taking

23 1

2. Model Rotation A second response paradigm on which younger children show considerable success at perspective taking is a model-rotation task. Children are asked to turn a model or duplicate of the array to indicate what would be seen from a specified position (or, in a study by Fishbein er al., 1972, to show to a person who is in a certain position a view of the display, as shown in a picture). Borke (1975) showed that children as young as 3 or 4 years performed surprisingly well on this task, even achieving 42% and 67% correct responses respectively, on a three-mountains display. Borke did not specifically compare these data to those obtained with another response mode, but the superiority of modelrotation to construction tasks has been shown by Rosser (1983) and Rosser et al. (1985), and the superiority of model rotation to picture-selection tasks has been shown by Fishbein er al. (1972) and Horan and Rosser (1983). From the point of view of spatial representation, the important aspect of model-rotation techniques is that the device itself preserves the internal spatial relationships among the objects in the array. Thus, the child needs to be able to infer only which object will be nearest an observer after a specified degree of movement. This inference they can evidently make. Piaget and Inhelder in fact noted the same phenomenon although not at ages as young as those studied by subsequent investigators. A question arises concerning why model rotation is easier than picture selection on this analysis, when most picture-selection tasks, including those of Fishbein et al. and Horan and Rosser, include only pictures of possible views of the array. That is, the pictures from which the child must choose do not include internal rearrangements of the elements, and successful choice is possible if the child can infer what object will be nearest the hypothetical observer. Several relevant factors, however, probably increase the difficulty of picture selection over model rotation. First, picture selection requires the child to scan and compare a set of similar pictures, something young children are known not to do efficiently or exhaustively (e.g., Vurpillot, 1968). By contrast, in model rotation, the child has to look only at a single array. Second, model rotation gives the child an opportunity to see the array transformed in a continuous fashion, which may help in imagining spatial transformation. By contrast, picture selection only shows the discrete static outcomes of spatial transformations. Third, the model-rotation task encourages the child to treat the task as one of array rotation rather than observer movement and the former has been shown to be easier than the latter even when picture selection is used for both (Huttenlocher & Presson, 1973). Fourth, the modelrotation device may allow the child to engage in more experimentation and trial and error, but the demand characteristics of the picture-selection task are such that a child may feel the need to point to a picture and stay committed to that answer. In summary, the fact that rotation devices preserve the

232

Nora Newcombe

internal structure of the array for the child is important from the spatial representation viewpoint, but several other aspects of the paradigm probably also contribute to children’s early success with it.

3. Picture Selection Picture-selection tasks are the most widely used technique for studying perspective taking. As suggested, however, they have demands of their own, some of which may increase the difficulty of showing the correct perspective. First, pictures are two-dimensional representations of a three-dimensional array. Nigl and Fishbein (1974, Experiment 3) have shown that offering a choice of three-dimensional models rather than pictures leads to better performance for both 6- and 10-year-olds. The advantage of models appeared for both the younger and the older children in conditions where ceiling and floor effects were avoided. A second fact about picture selection is that it requires the scanning and comparison of similar pictures, something younger children might be expected to find more difficult than older ones. In fact, increasing the number of choices seems to make the task harder, although this effect may not be differentially marked for younger subjects as one might expect. Fishbein et al. (1972) studied the ways that having four or eight response choices affected 4; 6- and 8-yearolds. They found that increasing the number of choices decreased accuracy, but that this effect was as marked for the older as for the younger children. Salatas and Flavell (1976) found no difference in number correct between a seven-item and a nine-item answer set, for 6- and 8-year-olds, but they did find more egocentric errors for the larger set. Age differences in this tendency were not reported. Thus, research to date indicates that increasing the number of pictures to be examined decreases performance in the age range from 4 to 8 years. No age-related interactions have emerged, however, suggesting that a greater number of choices is a special problem for the younger subjects, one which might mask their true spatial abilities. However, more research, using a greater variety of ages and numbers of pictures might show such an effect. A third fact about picture selection, discussed by Huttenlocher and Presson (1979), is that the selection must be made with the pictures in the same room as the array, and thus with the items in the array shown in incorrect relationship to the landmarks in the room. If the items have been encoded using the landmarks as suggested by the work of Presson (1980),this kind of incorrect relationship might be confusing. A last aspect of most picture-selection tasks is one that should make the task easier rather than harder. Many researchers present as choices only views of the array that are actually possible. That is, they do not show views of scrambled or rearranged arrays. Thus success is possible if children can compute the relationship of only one object to the observer correctly, perhaps

Deselopment of Spatial Perspective Taking

233

what object would be nearest the observer, as originally suggested by Piaget and Inhelder. Work by Cox (1978a, 1978b) and Bialystock (1986) indicates the importance of this factor. These investigators included among the response choices pictures showing the nearest object to the observer correctly, but other objects in rearranged order. Children up to the age of 9 or 10 years were quite likely to choose the scrambled picture. Thus, picture-selection tasks preserving the internal structure of the array probably overestimate the ability of younger children to solve spatial transformation problems in that they allow for success on the basis of ability to compute which object would be nearest an observer, without a need to transform the positions of the other objects. The data also indicate that at least until the age of 9 or 10 years, children do not encode the internal structure of the array, but rather consider the position of each object separately. Adults may also fail to encode the internal structure of the array (Presson, 1982) but what they may do, unlike younger subjects, is to check additional objects to see if the picture they have chosen is the correct answer.

4. Model Building Piaget and Inhelder reported similar results for model building as for picture selection. Subsequent work has shown that building appears to be harder overall than picture selection (Hoy, 1974) as well as harder than model rotation (Rosser, 1983); but has confirmed that very similar patterns of results appear for model building and picture selection (Hoy, 1974; Presson, 1982). Model building has two advantages as a dependent variable. One is that it is helpful in studies of adults who make few errors on other tasks (Presson, 1982). A second is that it allows inferences about mental processes to be made from examination of the order of placement of the elements (P. L. Harris & Bassett, 1976). In particular, Harris and Bassett noted that children as young as 4 years generally position first the element of an array that would be closest to an imagined observer even when they then continue to build the array so that the finished product shows an egocentric view. This ordering suggests that subjects as young as 4 years can imagine the position of the other and compute what object would be closest to the observer’s position. They may have difficulty, however, in working out the relationships of any other elements to the observer and thus fall back on an egocentric response. Alternatively, they may be overwhelmed as they build by the fact that they are asked to position targets within the same framework of external landmarks with respect to which they coded the original positions.

5, Item and Ibsition Questions Huttenlocher and Presson (1979) contrasted the performance of 8-year-old children on a model-selection task (appearance question) with their ability

234

Nora Newcornbe

to answer questions about which item would be in a specified position after a specified degree of observer movement (item question). That is, on the item question, children were asked which of four objects would be close to them (or far away, or on their 1eftA‘red” side or right2‘green” side) if they were seated at other positions around the array. Answers to these questions were 80% correct, and answers to appearance questions were only 44% correct. The better performance was not due to ease of understanding the question per se, because when children were asked to imagine the array moving on its own axis, rather than the observer moving, item questions were actually harder than appearance questions (see below). Nor was the ease of item questions due to their focus on one element of the array at a time, because position questions (if a viewer were at a specified point, where would a particular object be?) are much harder (Hardwick, Mclntyre, & Pick, 1976; Presson, 1982) The best way to explain the ease of item questions seems to be that the locations of items in the array are coded individually with reference to the external landmarks in the room. When the hypothetical observer is also located with respect to this framework, the relations between objects and the observer can be read from this representation. In contrast, selecting a model to show the observer’s view requires the child to ignore a model (the egocentric choice) showing the targets in their correct relationship to the external framework, in favor of a model showing the targets in an incorrect relationship to this framework. Presson (1987) presents an extended discussion of data of this kind, using the concepts of primary and secondary uses of spatial information. He argues that picture selection requires the use of an abstract, secondary frame of reference rather than the primary frame of reference (the room). The ability to perform such secondary analysis is said to develop with age, rather than the abilities to encode objects or perform mental transformations. Thus, by the age of 8 or 9 years, children seem to code spatial location in a fashion sufficient to allow them to compute the location of objects relative to an observer in another position. Furthermore, the fact that adults show similar patterns of performance (Presson, 1982) suggests that no developmental change in spatial encoding occurs after this age. However, 8 or 9 years is close to the 9 or 10 years at which Piaget and Inhelder believed that children develop understanding of projective space. A crucial question thus becomes whether younger children also show good performance on item questions. Newcombe (1989) has demonstrated that 5-year-olds also do well on item questions while showing the usual difficulty with appearance questions. Even younger children may be able to answer item questions successfully. P. L. Harris and Bassett (1976) found that 4-year-olds in a model-building task place first the object nearest a hypothetical observer which implies the

Development of Spatial firspective Taking

235

ability to work out at least one observer-object relationship. Ives (1980) using a single object (house, horse) as a stimulus asked children to say whether they would see its front, back, or side from a specified position (a single-object analogue of an item question). Responses were about 90% correct even among 3-year-olds; picture selection, however, was much more difficult. How preschool children would perform in answering item questions concerning a multiple-object array is of course unknown.

E. SUMMARY

Of the many variables reviewed in the section, some do not appear to show a relationship to perspective-taking performance. These include shielding of the display; the relative difficulty of go", 180°, and 270" movement; the number of objects in the display (when position information must be processed); and near-far versus left-right (when interposition is avoided). Relationships of other variables to perspective-taking have not been clearly assessed. These include motivational context, clarity of instructions, and familiarity of objects. A relatively large number of factors has been identified that do make perspective taking easier or harder. However, not all bear on the issue of how children code location and whether this coding is different from that of adults. Children of age 6 years and younger have difficulty with displays involving occlusion, because they wish to show all objects they know to be present as being present. Children who are generally good at spatial analysis may solve perspective-taking problems more easily. Perspective taking is easier (or at least less likely to be egocentric) when people are used as others rather than dolls, when children are not asked to show their own view first, when children first play with the materials or work with them in groups, when children walk around the display before being asked questions, when children are given training, and when displays do not differ from answer choices in dimensionality. However, none of these findings tells us how children code location. Several results do, however, bear on location coding, arguing for the hypothesis that the locations of array elements are coded individually with respect to an external framework of landmarks. Thus, perspective tasks are easy, even at 2 years, when children actually move around the array, because the targets can be located with respect to perceptually available landmarks. Similarly, perspective tasks are made easier by provision of a model of the room, showing landmarks as they would be located after movement to another position. The response children are asked to make is also important. Item questions (as opposed to position questions) can be answered well, at least by 5 years and possibly younger. This fact supports the idea that locations

236

Nora Newcombe

of targets and observers are coded with respect to the external framework and that perspective taking is possible when answers can be read from the representation. Model-rotation devices are easy because the internal relationships of the display are maintained, and children can succeed by locating one object correctly. Most likely, children code the relationship of the observer to the object directly in front, and move the response device until that object is in front of them. This ability to code observer-near-object relationships is also shown by Harris and Bassett's finding that children as young as 4 years locate that object first in a model-building task. Overall, the impression gained from these findings is that location coding with respect to external frameworks is established at ages considerably younger than 9 or 10 years.

V. Related Tasks A. EUCLIDEAN SPACE

The three-mountains task is described in 1 of the 15 chapters of The Child's Conception ofSpace. Many other experiments assessing understanding of projective space are also presented (e.g., construction of a straight line, projection of shadows). But little work has been done, except by Laurendeau and Pinard (1970), to examine these other techniques or the relationship of success on them to success on the three-mountains task. Of the five chapters dealing with Euclidean space, two chapters have received more attention: work on children's ability to reproduce a model oriented differently from the child's model, and work on children's understanding of horizontality and verticality. Within the context of an interest in perspective taking, two questions about these tasks are of central importance. First, do children succeed on them at about the same time as on the three-mountains task, as might be expected if projective and Euclidean space are closely related? Second, can children's problems with these tasks be linked to the hypothesis advanced in this article concerning what makes various versions of perspective taking harder or easier? That is, would one have problems with these tasks if the location of objects is coded individually with respect to an external framework? 1. Reproducing Model Placements

Piaget and Inhelder presented to children a model of a village, showing a stream, a road crossing, hills, trees, and houses. In one task, children were asked to place a doll on a second model, rotated 180" and separated from the first by a screen. At 3 or 4 years, placements were determined by closeness

Development of Spatial Rmpective Taking

237

to a single salient object or background. From 4 to 7 years, children exhibited some ability to place the doll correctly; by 7 to 8 years, placements were determined by logical multiplication of left-right and before-behind relationships. Success occurred later on a more complex task, that of reproducing the whole model either diagrammatically or by using model materials. In this task, a system of reference was established by ages 7 to 10 years, with distance and proportion not fully accurate until after that. Using model villages and the doll-placement task, Laurendeau and Pinard (1970) obtained similar results, with placements by younger children determined with reference to the self or a single point in the model. Laurendeau and Pinard did not find complete success on the most difficult placements until 10 years of age. Using more abstract materials and no topographic cues, Pufall and Shaw (1973) found that 10-year-olds still had difficulty in accurate placement. However, Pufall (1975) found better performance in 5-year-olds who used a farm scene; the children also did better when the child's model was rotated 90" rather than 180" from the experimenter's. Russell (1982) found that 5-year-olds improved somewhat on an allocentric placement task following training, but their performance was still quite poor. These studies show a rough congruence between ages of success on perspective taking and allocentric placement tasks. But of course, the same could be said for perspective taking and conceptually unrelated cognitive achievements-for example, use of memory strategies. Laurendeau and Pinard (1970) and De Lisi et al. (1976) studied perspective taking and allocentric placement in children ages 4 to 12 and 6 to 11 years old in the respective studies, and found that success on the two tasks was related. However, with age varying as widely as it did, this association does not demonstrate that the two tasks are indexing acquisition of related systems of spatial representation, as Laurendeau and Pinard recognized. Laurendeau and Pinard showed that a scalogram analysis of their five spatial tests indicated a consistent order of acquisition across subjects, of the substages of thought delineated for each of the tests separately. They argued that this finding indicated that a common line of development was tapped by all five tests, because inclusion of sequences from tests of causal thinking lowered the scalability of the data. However, the lowering was not large and it was not tested for statistical significance. General mental development seems adequate to explain the scalogram results. The approach to spatial coding and the perspective-taking task taken in this article suggests the hypothesis that allocentric placement is difficult for the same reason that perspective taking with appearance questions is difficult: The child is asked to place objects in manifestly different positions with respect to the external framework of the room. This hypothesis would explain the substantially greater difficulty of allocentric placement in the model studies

238

Nora Newcornbe

as compared to the performance levels found in studies of “large-scale space” (e.g., Herman, Roth, Miranda, & Getz, 1982). Herman et al. found that children as young as 5 years had little difficulty in reproducing a layout of eight unrelated objects after moving to the other side of the room. In this case, the physical relationship of the targets to the external framework remains invariant, Success with a rotated model however, requires the suppression of attention to external cues and a focus instead on the internal relationships among objects in the model. Such internal coding is a difficult task, as discussed further in the next subsection.

2. Horizontality and Verticality Considerable work has been done on children’s understanding of orientation to the gravitationally defined horizontal and vertical-for example, as in understanding of the fact that still water levels are horizontal (e.g., De Lisi, 1983; Liben, 1975; Thomas & Jamison, 1975). Much of this work has concerned the persistence of errors on these tasks into adulthood, especially among women (e.g., Liben & Golbeck, 1980). However, the majority of subjects are doing well by early adolescence. Thus, again, some rough congruence exists with perspective taking; although adults rarely make errors on the picture-choice task, they d o have some difficulty with building models (Presson, 1982). Horizontality-verticality and perspective-taking tasks have been given to the same groups of subjects by Larsen and Abravenel (1972). They found significantly earlier acquisition of horizontality-verticality than of perspective taking among 5 - to 10-year-olds; they did not assess correlations among the tasks. De Lisi et al. (1976) showed a correlation between horizontalityverticality and success in allocentric placement among 6- to 11-year-olds. In a potentially more stringent test of the relatedness of projective and Euclidean concepts, Cox (1977~)found no transfer of successful training of 5-yearolds in perspective-taking to water-level task. The difficulty of horizontality-verticality tasks seems closely related to the problem of selecting the correct framework within which to draw the stimulus. Children often choose the sides of the bottle rather than the gravitational horizontal as a referent for drawing water level or the sides of the hill rather than the gravitational vertical for drawing plumb lines. Their problems thus d o not seem to involve use of external landmarks in the room, as d o problems with perspective taking. This fact is not surprising, given that the stimuli are self-contained flat pictures rather than a group of objects with physical locations in the room. B. MENTAL ROTATION

Piaget and Inhelder (1966A971) reported the results of extensive investigations of children’s ability to imagine static and moving objects, both already

De\lelopment of Spatial FPrspectiiv Taking

239

seen and anticipated. One of their main arguments was that “at about 7 to 8 years a capacity for imaginal anticipation makes its first appearance, enabling the subject to reconstitute kinetic or transformational processes, and even foresee other simple sequences” (p. 358). The lack of imagery in preoperational thought was linked to the general tendency of thought at that stage to focus on states rather than transformations. By contrast, operational thought was said to allow the coordination of successive states and the transformations that lead from one to the other. The finding that children could imagine the outcome of rotation of an object on its own axis by the age of 7 or 8 years contrasts somewhat with the conclusion that perspective taking cannot be carried out until 9 or 10 years, but Piaget and Inhelder did not directly compare their results on mental rotation to results on perspective taking. However, Huttenlocher and Presson (1973) showed that for 8- and 10-year-olds, imagining the outcome of mental rotation of an array on its axis was in fact easier than imagining what an array would look like to a hypothetical observer using a model-selection task (appearance questions). These findings have been replicated by Finlay (1977) with 7- and 8- as well as 9- and I0-year-olds. P. L. Harris and Bassett (1976) also found rotation easier than perspective taking. Array rotation is usually easier than perspective taking, in that it does not require the child to locate a hypothetical observer before transforming the array. Thus, solving these problems requires fewer mental steps (Huttenlocher & Presson, 1973), as long as the response allows the subject to focus on only one object (appearance questions) or on one object at a time (model building). Array rotation is difficult when answering item questions, however, because the subject must transform all the objects in the array to decide which one occupies the specified position. Such a task would not be hard if the locations of the objects were internally coded with respect to each other; the difficulty of array rotation with item questions in contrast to its ease with appearance questions is what suggests that such internal coding does not occur even in adults (Presson, 1982). Thus, mental rotation seems to be either easier or harder than perspective taking, depending on whether the dependent variable used places a premium on internal coding for rotation problems (as item questions do) or on abstracting coding away from an external framework (as appearance questions do). For perspective taking, data suggest that as young as 5 years, children are able to perform effectively as long as they do not need to abstract away from an external framework, with item questions (Newcombe, 1989). Can mental rotation be observed at ages younger than the 7 or 8 years suggested by Piaget and Inhelder? Considerable controversy concerns whether mental rotation is possible in children as young as 4 or 5 years. Much of the research concerns rotation of single objects. Marmor (1975, 1977) presented evidence that children as

240

Nora Newcombe

young as 4 or 5 years showed linear slopes of reaction time as a function of the amount of rotation required, suggesting that they did perform mental rotation, Other investigators have failed to find linear slopes for younger subjects (e.g., Dean & Harvey, 1979). In a 1987 review, Dubas concluded that although children as young as 4 or 5 year often do show linear slopes, many of them may be cognitively precocious. Error rates are high at these ages. Dubas also found that when concrete operations are tested, demonstrably preoperational children do not show reliable evidence of linear slopes. These children may be performing rotation tasks, when they can do so at all, by memorizing the appearance of the stimulus at various degrees of rotation. The evidence on mental rotation of multiple-object arrays by young children is quite sparse. Using a multiple-location hiding task, Lasky, Romano, and Wenters (1980) found that children of 3,4, and 5 years were at or below chance. Performance was better than chance at 7 years, but not really good until age 10. In summary, current evidence suggests that array rotation is neither easier nor harder, across the board, than perspective taking. The difficulty of each depends on the exact coding requirements of the task. Very young children have great difficulty with internal coding or separated targets and with abstracting codings with respect to external landmarks away from these landmarks. But older children and adults continue to have marked difficulty with these same processes.

VI. Conclusion Children’s solutions to spatial problems pose an intriguing puzzle in development. For some time, difficulties with such tasks have been thought to index a pervasive characteristic of children’s thought, called “egocentrism,” but recent research has shown clearly that at least by 2 years of age, children are not egocentric in the sense of not knowing that other people see displays differently. The main argument of this article has been that perspective-taking tasks, and other spatial problems such as mental rotation, are difficult (or easy) for reasons that include the demands the particular problem makes on the usual form of representation of space. From at least 5 years of age, and perhaps earlier, this representation seems to be one in which small movable targets are encoded in relation to a framework of fixed landmarks, rather than in relation to each other. Such coding makes ecological sense, because movable items, almost by definition, make poor reference points for the location of other objects. Thus, spatial coding does not seem to change across middle childhood in the fashion discussed by Piaget. What remains for the future is to explore whether frameworks of landmarks are used soon after the beginning of free locomotion, and how location coding changes, if it does, in the first few years of life.

Development of Spatial Brspective Taking

241

From the point of view of investigators primarily interested in perspective taking, the main conclusion of this chapter is that an intellectually respectable form of perspective taking can be demonstrated as early as 5 years. Many factors can add to the difficulty of the task for children in middle childhood, and some of these performance factors are of interest in their own right. But investigation of them should not proceed as if their existence precluded the existence of a perspective-taking competence.

ACKNOWLEDGMENTS I thank Rochel Gelman and the psychology department at The University of Pennsylvania for providing a congenial atmosphere for working on this article, and Temple University for giving me a study leave. 1 also thank Janellen Huttenlocher and Clark Presson for reading the article, and Carolyn Spies for help with the literature search.

REFERENCES Acredolo, L. P. (1979). Laboratory versus home: The effect of environment on the 9-month-old infants’s choice of spatial reference system. Developmental Psychology, 15, 666-667. Acredolo, L. P., &Evans, D. (1980). Developmental changes in the effects of landmarks on infant spatial behavior. Developmental Psychology, 16, 312-318. Acredolo, L. P., Pick, H. L., & Olsen, M. G. (1975). Environmental differentiation and familiarity as determinants of children’s memory for spatial location. Developmental Psychology, 11, 495-501. Aebli, H. (1967). Egocentrism (Piaget) not a phase of mental development but a ’substitute solution’ for an insoluble task. Bdagogica Europaea, 3, 97-103. Bearison, D. J., Magzamen, S., & Filardo, E. K. (1986). Socio-cognitive conflict and cognitive growth in young children. Merrill-thlmer Quarterly, 32, 51-72. Bialystok, E. (1986, August). Childrenk mental transformations of spatial displays. Paper presented at the meeting of the American Psychological Association, Washington, DC. Borke, H. (1975). Piaget’s mountains revisited: Changes in the egocentric landscape. Developmental Psychology, 11, 240-243. Bowd, A. (1975). The relationship between perceptual egocentrism and field dependence in early childhood. Journal of Genetic Psychology, 121, 63-69. Braine, L., & Eder, R. (1983). Left-right memory in 2-year-old children: A new look at search tasks. Psychology, 19, 45-55. Bremner, 3 . C . (1978). Egocentric versus allocentric spatial coding in nine-month-old infants: Factors influencing the choice of code. Developmental Psychology, 14, 346-355. Brodzinsky, D. M. (1980). Cognitive style differences in children’s spatial perspective taking. Developmental Psychology, 16, 151-152. Brodzinsky, D. M. (1982). Relationship between cognitive style and cognitive development: A 2-year longitudinal study. Developmental Psychology, 18, 617-626. Brodzinsky, D. M., Jackson, J. P., & Overton, W. F. (1972). Effects of perceptual shielding in the development of spatial perspectives. Child Development, 43, 1041-1046. Bryant, P. E., & Somerville, S. C. (1986). The spatial demands of graphs. British Journal of &ChOlOgy, 11, 187-197.

242

Nora Newcombe

Coie, J. D., Costanzo, P. R., & Farnill, D. (1973). Specific transitions in the development of spatial perspective-taking ability. Developmental Psychology, 9, 167-1 77. Cox, M. V. (1975). The other observer in a perspectives task. British Journal of Educational PSyChOlOgy, 45, 83-85. Cox, M. V. (1977a). Perspective ability: The other observer in the task. Perceptual and Motor Skills, 44, 76. Cox, M. V. (1977b). Perspective ability: The relative difficulty of the other observer’s viewpoints. Journal of Experimental Child Psychology. 24, 254-259. Cox, M. V. (1977~).Teaching perspective ability to five-year-olds. British Journal of Educational PSyChOlOgj, 47, 312-321. Cox, M. V. (1978a). The development of perspective-takingability in children. International Journal of Behavioral Development, 1, 247-254. Cox, M. V. (1978b). Order of the acquisition of perspective-takingskills. helopmentalPsychology, 14, 421-422. Cox, M. V., & Willetts, E. (1982). Childhood egocentrism: The order of acquisition of beforebehind and left-right relationships. British Journal of Educational Psychology, 52, 366-369. Dean, A. L., & Harvey, W. 0. (1979). An information processing analysis of a Piagetian imagery task. Developmenla1 Psychology, 15, 474-476. De Lisi, R. (1983). Developmental and individual differences in children’s representation of the horizontal coordinate. Merrill-&her Quarterly, 29, 179-196. De Lisi, R., Locker, R., & Youniss, J. (1976). Anticipatory imagery and spatial operations. Developmental Psychology, 12, 298-3 10. Doise, W., Mugny, G., & Perret-Clermont, A.-N. (1975). Social interaction and the development of cognitive operations. European Journal of Social Psychology, 5 , 367-383. Donaldson, M. (1978). Childrenk minds. New York: Norton. Dubas, J. S. (1987, April). Developmental changes in performance on the mental rotation task: A meta-onalysis. Paper presented at the meeting of the Society for Research in Child Development, Baltimore, MD. Eiser, C. (1974). Recognition and inference in the coordination of perspectives. British Journal of Educational Psychology, 44, 309-312. Eiser, C. (1976). Questions children ask about spatial arrays: An analysis of the processes involved in coordinating perspectives. British Journal of Education Psychology, 46, 203-211. Eliot, J., & Dayton, C. M. (1976). Factors affecting accuracy of perception on a task requiring the ability to identify viewpoints. Journal of Genetic Psychologj, 128, 201-214. Epstein, S. (1979). The stability of behavior: I. On predicting most of the people much of the time. Journal of Personality and Social Psychologj, 37, 1097-1126. Fehr, L. A. (1978). Methodological inconsistencies in the measurement of spatial perspectivetaking ability: A cause for concern. Human Development, 21, 302-315. Fehr, L. A. (1979). Hypotheticality and the other observer in a perspective task. British Journal of Educational Psychology, 49, 93-96. Fehr, L. A. (1980). Spatial landmarks revisited: Are they useful? Journal of Genetic Psychology, 136, 299-300. Fehr, L. A., & Fishbein, H. D. (1976). The effects of an explicit landmark on spatial judgment. In P. Suedfeld & J. A. Russell (Eds.), The behavioralbasisof design (pp. 86-93). Stroudsburg, PA: Dowden, Hutchinson & Ross. Fehr, L. A., Lapsley, D. K., Enright, R. D., McMahon, P. M., & Ackerman. A. M. (1983). Coordination of perspectives: The importance of stimulus dimensionality. Journal of General PSyChOlO~,108, 73-78. Fehr, L. A., McMahon, P. M., & Fehr, K. M. (1982). Coordination of perspectives: Orientations of the other observer. Journal of Genetic Psychology, 140, 151-152.

Development of Spatial Arspective Taking

243

Finlay, D. C. (1977). Rotation and perspective tasks in 7-8 and 9-10-year-old children. Brceptual and Motor Skills, 44, 1216-1218. Finley, G. E., Solla, J., & Cowan, P. A. (1977). Field dependence-independence, egocentrism, and conservation in young children. Journal of Genetic Psychology, 131, 155-156. Fishbein, H.. Lewis, S., & Keiffer, K. (1972). Children’s understanding of spatial relations: Coordination of perspectives. Developmental Psychology, 7, 21-33. Flavell, J. H. (1971). Stage-related properties of cognitive development. Cognitive Psychology, 2, 421-453. Flavell, J. H., Botkin, P. T., Fry, C. L., Wright, J. W., & Jarvis, P. E. (1968). Thedevelopment of role-taking and communication skills in children. New York: Wiley. Flavell, J. H., Everett, B. A,, Croft, K., & Flavell, E. R. (1981). Young children’s knowledge about visual perception: Further evidence for the Level 1-Level 2 distinction. Developmental P~ychology,17, 99-103. Flavell, J. H., Flavell, E. R., Green, F. L., & Wilcox, S. A. (1980). Young children’s knowledge about unusual perception: Effect of observer’s distance from target on perceptional clarity of target. Developmental Psychology, 16, 10-12. Flavell, J. H., Flavell, E. R., Green, F. L., & Wilcox, S. A. (1981). The development of three spatial perspective-taking rules. Child Development. 52, 356-358. Flavell, J . H., Omanson. R. C., & Latham, C. (1978). Solving spatial perspective-taking problems by rule vs. computation: A developmental study. Developmental Psychology, 14 462-473. Flavell, J . H., Shipstead, S. G., & Croft, K. (1978). Young children’s knowledge about visual perception: Hiding objects from others. Child Development, 49, 1208-1211. Flavell, J. H., Shipstead, S. G., & Croft, K . (1980). What young children think you see when their eyes are closed. Cognition, 8, 369-387. Ford, M. E. (1979). The construct validity of egocentrism. Psychological Bulletin, 86, 1169-1188. Ford, M. E. (1985). Two perspectives on the validation of developmental constructs: Psychometric and theoretical limitations in research on egocentrism. Psychological Bulletin. 97, 497-501. Garner, J., & Plant, E. L. (1972). On the measurement of egocentrism: A replication and extension of Aebli’s findings. British Journal of Educational Psychology, 42, 79-83. Gelman, R., & Baillargeon, R. (1983). A review of some Piagetian concepts. In P. H. Mussen, J. H. Flavell, & E. M. Markman (Eds.), Handbook ofchild psychology: Vol. 3. Cognitive development (pp. 167-230). New York: Wiley. Gullo, D. F., & Bersani, C. (1983). Effects of three experimental conditions on preschool children’s ability to coordinate visual perspectives. Arceptual and Motor Skills, 56, 675-678. Gzesh, S. M., & Surber, C. F. (1985). Visual perspective-taking skills in children. ChildDevelop ment. 56, 1204-1213. Hardwick, D., McIntyre, C., & Pick, H. (1976). The content and manipulation of cognitive maps in children and adults. Monographs of the Societyfor Research in Child Development, 41 (3, Serial No. 166). Harris, L. J., & Strommen, E . A. (1972). The role of front-back features in children’s front, back and beside placements of objects. Merrill-&her Quarterly, 18, 259-271. Harris, P. L., & Bassett, E. (1976). Reconstruction from the mental image. Journalof Experimental Child Psychology, 21, 514-523. Herman, J. F., Roth, S. F., Miranda, C., & Getz, M. (1982). Children’s memory for spatial locations: The influence of recall perspective and type of environment. Journal of Expermen[a1 Child Psychology, 34, 257-273. Herman, J. F., & Siegel, A. W. (1978). The development of cognitive mapping of the large-scale environment. Journal of Experimental Child Psychology. 26, 389-406. Hobson, R. P. (1980). The question of egocentrism: The young child’s competence in the

Nora Newcombe

244

coordination of perspectives. Journal of Child Psychology, and Psychiatry and Allied

Discipline, 21, 325-33 1. Hobson, R. P. (1982). The question of childhood egocentrism: The coordination of perspectives in relation to operational thinking. Journal of Child Psychology and Psychiatry and Allied Disciplines, 23, 43-60. Horan, P. F., & Rosser, R. A. (1983). The function of the response mode in the coordination of perspectives. Contemporary Educational Psychology, 8, 347-354. Hoy, E. A. (1974). Predicting another’s visual perspectives. Developmentol Psychofogy, 10, 462. Hughes, M. (1978). Selecting pictures of another person’s view. British Journal of Educational f%yChOlOgy,41, 210-219. Hughes, M., & Donaldson, M. (1979). The use of hiding games for studying the coordination of viewpoints. Educational Review, 31, 133-140. Huttenlocher, J., & Newcombe, N. (1984). The child‘s representation of information about location. In C. Sophian (Ed.), The origin of cognitive skills (pp. 81-111). Hillsdale, NJ: Erlbaum. Huttenlocher, J., & Presson, C. C. (1973). Mental rotation and the perspective problem. Cognitive Psychology, 4, 277-299. Huttenlocher, J., & Presson, C. C. (1979). The coding and transformation of spatial information. Cognitive Psychology, 11, 375-394. Ives, W. (1980). Preschool children’s ability to coordinate spatial perspectives through language and pictures. Child Development, 51, 1303-1306. Jacobsen, T. L., & Waters, H. S. (1985). Spatial perspective taking: Coordination of left-right and near-far spatial dimensions. Journal of Experimental Child Psychology, 39, 72-84. Kielgast, K. (1971). Piaget’s concept of spatial egocentrism: A reevaluation. Scandinavian Journal Of Psychology, 12, 79-91. Knudson, K. H., & Kagan, S. (1977). Visual perspective role-taking and field-independenceamong Anglo-American and Mexican-Americanchildren of two ages. Journal of Genetic Psychology, 131, 243-253. Landau, B. (1988). The construction and use of spatial knowledge in blind and sighted children. In J. Stiles-Davis, M. Kritchevsky, & U. Bellugi (Eds.), Spatial cognifion: Brain bases and development (pp. 343-371). Hillsdale, NJ: Erlbaum. Landau, B., Gleitman, H., & Spelke, E. (1981). Spatial knowledge and geometric representation in a child blind from birth. Science, 213, 1275-1278. Lapsley, D. K., Fehr, L. A., & Enright, R. D. (1981). Coordination of perspectives: The comparability of results. Journal of Genetic Psychology, 138, 311-312. Larsen, G. Y., & Abravanel, E. (1972). An examination of the developmental relations between certain spatial tasks. Merrill-hlmer Quarterly, 18, 39-51. Lasky, R. E., Romano, N., & Wenters, J. (1980). Spatial localization in children after changes in position. Journal of Experimental Child Psychology, 29, 225-248. Laurendeau, M., & Pinard, A. (1970). The development of the concept ofspace in the child. New York: International Universities Press. Lempers, J. D., Flavell, J. H., & Flavell, E. R. (1977). Development in very young children of tacit knowledge concerning visual perception. Genetic Psychology Monographs, 95, 3-53. Levine, S. C., & Carey, S. (1982). Up front: The acquisition of a concept and a word. Journal of Child Language, 9, 645-657. Liben, L. S. (1975). Long-term memory for pictures related to serviation, horizontality and verticality concepts. Developmental Psychology, 11, 795-806. Liben, L. S. (1978). Perspective-taking skills in young children: Seeing the world through rosecolored glasses. Developmental Psychology, 14, 87-92. Liben, L. S. (1988). Conceptual issues in the development of spatial cognition. In J. Stiles-Davis, M. Kritchevsky, & U. Bellugi (Eds.), Spatial cognition: Brain bases and development (pp. 167-194). Hillsdale, NJ: Erlbaum.

Development of Spatial &rsp:ctive Taking

245

Liben, L. S., & Belknap, B. (1981). Intellectual realism: Implications for investigations of perceptual perspective taking in young children. Child Development, 52, 921-924. Liben, L. S., & Golbeck, S. L. (1980). Sex differences in performance on Piagetian spatial tasks: Differences in competence or performance? Child Development, 51, 594-597. Light, P., & Nix, C. (1983). Own view vs. good view in a perspective-taking task. ChildDevelop-

ment, 52, 480-483. Lovell, K. (1959). A follow-up study of some aspects of the work of Piaget and Inhelder on the child’s conception of space. British Journal of Educational Psychology, 29, 104-117. Mandler, J. M. (1983). Representation. In P. H. Mussen, J. H. Flavell, & E. M. Markman (Eds.), Handbook of childpsychology: Vol. 3. Cognitive Development @p. 420-494). New York: Wiley. Marmor, G. S. (1975). Development of kinetic images: When does the child first represent movement in mental images? Cognitive Psychology, 7, 548-559. Marmor, G. S. (1977). Mental rotation and number conservation: Are they related? Develap-

mental Psychology, 13, 320-325. Masangkay, Z. S., McCluskey, K. A., McIntyre, C. W., Sims-Knight, J., Vaughn, B., & Flavell, J. H. (1974). The early development of inferences about the visual percepts of others. Child Development, 45, 357-366. Matthews. W. S., Beebe, S., & Bopp, M. (1980). Spatial perspective-takingand pretend play. &nep

tual and Motor Skills, 51, 49-50. McKenzie, B. E., Day, R. H., & Ihsen, E. (1984). Localization of events in space: Young infants are not always egocentric. British Journal of Developmental Psychology, 1, 1-10. Miller, J. W. (1967). Measuring perspective ability. Journal of Geography, 66, 167-171. Miller, J. W., Boismier, J. D., &Hooks, J. (1969). Training in spatial conceptualization: Teacher directed activity, automated and combination programs. Journal of Experimental Education. 38, 87-92. Minnigerode, F. A., & Carey, R. N. (1974). The development of mechanisms underlying spatial perspectives. Child Development, 45, 496-498. Mischel, W. (1968). Fkrsonality and assessment. New York: Wiley. Mom, J. R. (1986). The construction of perkspectives: Piaget’s alternative to spatial egocentrism. International Journal of Behavioral Development. 10, 263-279. Mugny, G., & Doise, W. (1978). Socio-cognitive conflict and structure of individual and collective performances. European Journal of Social Psychology, 8, 181-192. Newcombe, N. (1988). The paradox of proximity in early spatial representation. British Journal of Developmental Psychology, 6 , 376-378. Newcombe, N. (1989, May). &rspective taking and spatial representation: The role of environmental coding. Paper presented at the meeting of the Jean Piaget Society, Philadelphia, PA. Newcombe, N., Dubas, J. S., & Spies, C. (1985, April). 7bo-year-olds’codingof spatial location. Paper presented at the meeting of the Society of Research in Child Development, Toronto, Canada. Nigl, A. J., & Fishbein, H. D. (1974). Perception and conception in coordination of perspectives. Developmental Psychology. 10, 858-866. Okonji, M. O.,&Olagbaiye, 0.0.(1975). Field dependence and thecoordination of perspectives.

Developmental Psychology, 11, 520. Piaget, J. (1926). The language and thought of the child. London: Routledge & Kegan Paul. (Original work published 1923) Piaget, J., & Inhelder, B. (1967). The childs conception of space (F. J. Langdon & J. L. Lunzer, trans). New York: Norton. (Original work published 1948) Piaget, J., & Inhelder, B. (1971). Mental imagery in the child (P. A. Chilton, trans.). New York: Basic Books. (Original work published 1966) Piaget, J.. Inhelder, B., & Szeminska, A. (1981). The chi& conception of geometry (E. A. Lunzer, trans.). New York: Norton. (Original work published 1948)

246

Nora Newcombe

Pillow, B. H., & Flavell, J. H. (1985). Intellectual realism: The role of children’s interpretations of pictures and perceptual verbs. Child Development, 56, 664-670. Pillow, B. H., & Flavell, J. H. (1986). Young children’s knowledge about visual perception: Projective size and shape. Child Development, 57, 125-135. Presson, C. C. (1980). Spatial egocentrism and the effect of an alternate frame of reference. Journal of Experimental Child Psychology, 29, 391-402. Presson, C. C. (1982). Strategies in spatial reasoning. Journal of Experimental Psychology: Learning, Memory and Cognition. 8, 253-251. Presson, C. C. (1987). The development of spatial cognition: Secondary uses of spatial information. In N. Eisenberg (Ed.), Contemporarytopics in developmentalpsychology(pp. 77-112). New York: Wiley. Presson, C. C., & Ihrig, L. H. (1982). Using matter as a spatial landmark: Evidence against egocentric coding in infancy. Developmental Psychofogy, 18, 699-703. Presson, C. C., & Sornerville, S. C. (1985). Beyond egocentrism: A new look at the beginnings of spatial representation. In H . M. Wellman (Ed.), Children’ssearching: The development of search skill and spatial representation (pp. 1-26). Hillsdale, NJ: Erlbaum. Pufall, P. B. (1975). Egocentrism in spatial thinking: It depends on your point of view. Developmental Psychology, 11, 297-303. Pufall, P. B., & Shaw, R. E. (1973). Analysis of the development of children’s spatial reference systems. Cognitive Psychology, 5, 151-175. Rieser, J. (1979). Reference systems and the spatial orientation of six-month-old infants. Child Development, 50, 1078- 1087. Rieser, J.. & Edwards, K. (1979, August). ChildrenS perception and thegeometricstA multidimensionalscalinganalysis. Paper presented at the meeting of the American Psychological Association, New York. Rosser, R. A. (1983). The emergence of spatial perspective taking: An information processing alternative to egocentrism. Child Development, 54, 660-668. Rosser, R. A., Ensing, S. S., Mazzeo, J., & Horan, P. F. (1985). Visual perspective taking in children: Further ramifications of a n information processing model. Journal of Generic Psychology, 146, 379-387. Rushton, J. P., Brainerd, C. J., & Pressley, M. (1983). Behavioral development and construct validity: The principle of aggregation. Psychological Bulletin, 94, 18-38. Russell, J. (1982). Facilitation of children’s allocentric placement by reducing task complexity and providing a verbal rule. Journal of Genetic Psychology, 141, 203-212. Salatas, H., & Flavell, J. H. (1976). Perspective taking: The development of two components of knowledge. Child Development, 47, 103-109. Schachter, D., & Collin, E. S. (1979). Spatial perspective taking in young children. Journal of Experimental Child Psycholoa, 21, 467-478. Schatzow, M. D., Kahane, D. C., & Youniss, J. (1980). The effects of movment on perspective taking and the coordination of perspectives. Developmental Psychology, 16, 582-587. Shantz, C. U.(1975). The development of social cognition. In E. M. Hetherington (Ed.), Review ofchild development research (Vol. 5, pp. 257-232). Chicago, IL: University of Chicago Press. Shantz, C. U.,&Watson, J. S. (1970). Assessment of spatial egocentrism through expectancy violation. Psychonomic Science. 18, 93-94. Shantz, C. U.,& Watson, J. S. (1971). Spatial abilities and spatial egocentrism in the young child. Child Development, 42, 171-181. Shepard, R. N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science, 171, 701-703. Silverman, I. W. (1986). Paining children in spatial perspective taking. Paper presented at the meeting of the conference on Human Development, Nashville, TN.

Development of Spatial Perspective Taking

247

SomerviUe, S. C., & Bryant, P. E. (1985). Young children’s use of spatial coordinates. Child Development* 56, 604-613. Thomas, H., & Jamison, W. (1975). On the acquisition of understanding that still water is horizontal. Merrill-A?lmer Quarterly, 21, 3 1-44. Verkozen, J. (1975). Egocentrism: Stage or state? Psychoanalytic Review, 62, 305-308. Vurpillot, E. (1968). The development of scanning strategies and their relation to visual differentiation. Journal of Experimental Child Psychology. 6, 632-650. Walker, L. D., & Gollin, E. S. (1977). Perspective role taking in young children. Journal of Experimentat Child Psychology, 24, 343-357. Waters, €1. S., & Tinsley, V. S. (1985). Evaluating the discriminant and convergent validity of developmental constructs: Another look at the concept of egocentrism. AychologicalBulletin, 91, 483-496.

This Page Intentionally Left Blank

DEVELOPMENTAL STUDIES OF ALERTNESS AND ENCODING EFFECTS OF STIMULUS REPETITION

Daniel W Smothergill DEPARTMENT OF PSYCHOLOGY SYRACUSE UNIVERSITY SYRACUSE, NEW YORK 13244

Alan G. Kraut AMERICAN PSYCHOLQGICAL SOCIETY WASHINGTON, D.C. 20003

I. INTRODUCTION 11. THE STIMULUS FAMILIARIZATION EFFECT 111. COMPONENTS OF ATTENTION

A. THE ORIENTATION REACTION AND COMPONENTS OF ATTENTION B . ALERTNESS DECREMENT AND ENCODING FACILITATION: A TWO-FACTOR THEORY IV. AGE, STIMULUS FAMILIARIZATION EFFECTS, AND LATENT INHIBITION

V. MECHANISMS OF FAMILIARIZATION EFFECTS A. DELAYED TESTING B. NOMINAL AND FUNCTIONAL STIMULI

VI. STIMULUS CHARACTERISTICS VII. FAMILIARIZATION AS A TOOL IN STUDYING READING ACQUISITION VIII. SUMMARY AND CONCLUSIONS REFERENCES

249 ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR, VOL. 22

Copyright 0 1989 by Academic Press. Inc. All rights o f reproduction in any form reserved.

250

Daniel

N Smothergill and Alan G. Kraut

I. Introduction The metaphor of children as “universal novices,” gradually and continuously acquiring expertise in a variety of specific knowledge domains, is one of the major guiding ideas in contemporary research in cognitive development (Flavell, 1985). Although the metaphor implicates novelty as a ubiquitous fact of children’s lives, since the late 1960s, relatively little research has been addressed at mechanisms mediating novelty and familiarity effects. Earlier, by contrast, the issue had attracted sufficient research to be reviewed as the lead article in the inaugural volume of this series (Cantor, 1963). Somewhat paradoxically, the use of experimental techniques involving differential response to novel and familiar stimuli has increased greatly since Cantor’s review. In the study of infants, especially, the use of such techniques has become widespread (e.g., Bertenthal, Profitt, Spetner, & Thomas, 1985; Cohen & Strauss, 1979; Kellman & Short, 1987). It is worth noting, however, that although references to habituation are not uncommon in this literature, the term is almost always used more in the empirical sense of decrement and recovery of the dependent variable of interest, than in the theoretical sense described, for example, by Thompson and Spencer (1966). With few exceptions (e.g., Ruff, 1986), the question of what habituates has received little attention in infant research. An early and important exception to the predominantly heuristic research use of novelty and familiarity was seen in the serial habituation hypothesis proposed by Jeffrey (1968). This hypothesis suggested that perceptual development occurs by means of habituation of the orienting reaction to the cues in a stimulus complex. Cues were assumed to be habituated to in order from greatest to least salience, and a neurological record of the activity was assumed to be retained in memory. In this way, it was proposed, both basic perceptual learning and the acquisition of more abstract concepts, such as object permanence and conservation, might be accounted for. The serial habituation hypothesis aroused a great deal of interest, largely because it suggested how the gulf between the then-disparate research fields of experimental child psychology and perceptual-cognitive development might be bridged. Although direct tests proved difficult and few (see Jeffrey, 1976; Miller, 1972), the hypothesis was clearly a factor in the dramatic rise in the use of stimulus repetition procedures of the following decades. Lying between the pragmatic use of stimulus repetition, novelty, and familiarity that characterizes infant research, and the theoretical formalism of the serial habituation hypothesis, a gap has existed in research directed at understanding the psychological processes underlying novelty and familiarity effects in children. Such research could lead to a better conceptual understanding of familiarization effects in infant research studies, as well as illuminate the cognitive development of a universal novice.

Alertness and Encoding Eflects in Children

25 1

This article summarizes and integrates a program of research that has had as its focus the hypothesis that stimulus repetition brings about changes in two components of attention originally identified by Posner and Boles (1971) as alertness and encoding. This research, carried out since the late 1970s, originated in an effort to account for an effect of stimulus repetition in young children that was surprisingly robust but not well understood. Over the years, the theoretical ideas developed in the initial research were found useful in studying a variety of other questions, ranging from the possibility of developmental change in response to stimulus repetition to the acquisition of reading automaticity. A summary of individual studies is presented in Table I.

11. The Stimulus Familiarization Effect Cantor and Cantor (1964) were interested in how the prior presentation of a stimulus might influence its effectiveness as a conditioned stimulus. They had 5-year-olds merely observe 40 2-second presentations of either a white light or a buzzer. Subjects then performed an operant task by quickly removing their hand from a start position and pulling a single lever whenever either the familiarized or nonfamiliarized stimulus occurred. A reinforcer (a marble) was dispensed after each response. They found that starting speed (time from stimulus onset to slight lever depression) was slower for the familiarized stimulus across all 8 blocks of 6 test trials that were run. This study was the first demonstration of what came to be called the stimulusfamiliarization 6fect (Bogartz & Witte, 1966). The basic effect was shortly replicated at least 8 times in a series of experiments conducted with young children in the Iowa laboratories. Cantor (1969) provided a comprehensive review of these studies. Among the more important findings were the following: (1) the effect occurs reliably in both simple- and choice-reaction-time paradigms; (2) the effect is probably not due to conditioned passivity, in that it also occurs when subjects respond to stimulus presentation during familiarization (Bogartz & Witte, 1966); (3) about 20 familiarization trials are required to produce the effect; and (4) the effect occurs when subjects respond either to stimulus onset or to stimulus offset (Witte & Cantor, 1967). Cantor (1969) pointed out that the overall pattern of results was consistent with theorizing on habituation of the orientation reaction. Particularly compelling for this explanation were findings from several studies indicating that the familiarization effect is either weak or absent during the first block of test trials, but then emerges and persists across the remaining blocks. Cantor noted that this pattern is congruent with the notion that a habituated orientation reaction initially recovers when invested with signal character at the beginning of testing, but then rehabituates relatively rapidly as testing continues.

Daniel U! Smothergill and Alan G. Kraut

252

TABLE I Summary of Experiments on Alertness and Encoding Effects of Stimulus Repetition"

Study Kraut (1976)

Experiment number

Age (YO

1

7&8

2

7&8

3

7&8

Kraut & Smothergill

Adults

Major results Responses slower to familiarized stimulus than to novel stimulus Familiarized stimulus less effective than novel stimulus as warning signal Responses faster to familiarized stimulus than to novel stimulus when preceded by neutral warning signal Responses slower to familiarized stimulus than to novel stimulus

(1978)

Adults

Smothergill & Kraut

1

5

1

I & 8; 11 & 12

(1980)

Kraut & Smothergill (1980)

7 & 8; 11 & 12

Kraut, Smothergill, & Farkas

Adults

(1981)

2

Adults

3

Adults

Responses slower to familiarized stimulus than to novel stimulus when testing delayed 15 or 30 min., opposite pattern when testing immediate Response speed to colored forms directly related to salience of dimension unchanged from familiarization; simple reaction time Responses faster to novel word from familiarized category than to novel word from nonfamiliarized category (taxonomic categories) Same as Experiment 1 (good-bad poles of semantic differential) Colors - Responses slower to familiarized stimulus than to novel stimulus Words-Responses faster to familiarized stimulus than to novel stimulus Familiarized word less effective warning signal than novel word Simple reaction time to stimulus onset Responses slower to familiarized stimulus than to novel stimulus for both colors and words

Alertness and Encoding Eflects in Children

253

TABLE I (continued)

Study

Smothergill & Kraut

Experiment number

Age (yr)

4

Adults

1

Adults

1

Adults

(1981)

Cecil, Kraut, & Smothergill (1984)

Kraut & Smothergill

1

6&1

(readers); 10 & 11

(1986) 2

6&7

(non-readers) 3

6&7

(readers) 4

6&7

(readers)

Major results Simple reaction time to stimulus offset Same result as Experiment 3 for colors Null effect for familiarized and novel words Patterns as stimuli; response speed directly related to nearness of novel test stimulus to prototype inferred from familiarization Warning signal 0, 450, or 2500 msec before familiarized or novel stimulus Responses faster to familiarized stimulus than to novel stimulus at 450 msec; opposite pattern at 0 and 2500 milliseconds Responses slower to familiarized words than to novel words in younger group; opposite pattern for older group Responses faster to familiarized words than to novel words Warning signal did not change result obtained in Experiment 1 Responses slower to novel word from familiarized category than to novel word from nonfamiliarized category

OStimuli were projected slides and, except where noted, were colors.

Finally, a single attempt in the early studies to demonstrate the stimulus familiarization effect in adults proved unsuccessful (Meyers & Joseph, 1968). The possibility that the effect is either weak or absent in adults was also suggested by results obtained by an independent research team using a quite different paradigm (Lubow, Alek, & Arzy, 1975). The issue of age differences in familiarization effects has proven controversial and is discussed in Section IV.

254

Daniel K Smothergill and Alan G. Kraut

111. Components of Attention Attention reemerged as a popular topic in experimental psychology in the 1960s and 1970s. Its revival after decades of neglect was due to attempts to deal with practical problems, such as vigilance (Broadbent, 1958), and also to a sense of dissatisfaction with the ability of stimulus-response models to account for a growing number of experimental findings. In retrospect, attention provided a wedge that opened up a number of areas of research that had languished throughout the heyday of behaviorism. Posner and Boies (1971) presented one of the major models of attention to emerge during this period. They proposed that attention consists of three largely independent components referred to as alertness; set, or encoding and limited capacity. Because only the first two of these components are relevant for present purposes, the discussion that follows is limited to them. Alertness was conceptualized as a subcortical process that prepares the system in a general way to respond to incoming information. The nature of the incoming information was thought to be largely irrelevant to the alertness process. That is, activation of alertness was presumed to have a nonselective effect on the information processing system. Encoding, the second component of attention, referred to the internal coding of incoming information. By definition, encoding was selective and permitted same-different judgments to be performed. Posner and Boies reported a series of experiments supporting the independence of alertness and encoding. Essentially, they demonstrated in these experiments that preparation time (interval between warning signal and first target) and time to encode the first target (interval between targets) have additive effects on speed of same-different judgments for three different kinds of stimuli. A. THE ORIENTATION REACTION AND COMPONENTS OF ATTENTION

Certain similarities are apparent between the theoretical concept of orientation reaction on the one hand and the components-of-attention model on the other. From the orientation-reaction perspective, a novel stimulus elicits orienting; in the components-of-attention model, a warning signal increases alertness. Moreover, some part of an orientation reaction must be stimulus specific because repeated stimulus presentation is held to decrease the magnitude of the orientation reaction to that particular stimulus. With respect to the components-of-attention model, encoding is also regarded as stimulus specific. We noted in Section I1 that Cantor had mentioned that the orientation reaction might provide an understanding of the stimulus familiarization effect. Actually, Cantor expressed some reservations about such an understanding

Alertness and Encoding Effects in Children

255

on grounds that the orientation reaction concept seemed too poorly specified; he apparently pursued the idea no further at the time. The convergence between the orientation reaction concept and the alertness-encoding model of attention suggested, however, that the idea might now be profitably taken up in renewed form. Specifically, advantage could be taken of methodology developed in studying components of attention to determine whether the processes of alertness and encoding might provide a framework for understanding the stimulus familiarization effect. This, in brief, provided the rationale for a series of experiments Alan Kraut carried out as his master’s thesis research at Syracuse University. B. ALERTNESS DECREMENT AND ENCODING FACILITATION: A TWO-FACTOR THEORY

Kraut (1976) performed three experiments with 7- and 8-year-old children. The initial part of each experiment was the same. Subjects merely observed 30 3-second presentations of a slide of a simple colored form. Immediately afterward, a choice-reaction-time task was performed. In the first experiment, which was a replication of the stimulus familiarization effect, the familiar stimulus and a novel stimulus (another colored form) were used as target stimuli. As expected, responses to the familiar stimulus were found to be slower than those to the novel stimulus. Experiment 2 was designed to demonstrate that a decrement in alertness was responsible for the decrease in response speed to a familiarized stimulus. Subjects in this experiment responded to neutral targets: an inverted T and a cross. Just prior to target onset, either the familiarized stimulus or a novel stimulus was presented as a warning signal. The pairings of warning signals and targets were haphazard across trials; the warning signal that occurred on any given trial had essentially a .5 probability of being followed by either target. The hypothesis that diminished response speed to a familiarized stimulus is due to a decrement in alertness was supported by the finding that slower responses occurred on trials on which the familiarized stimulus was the warning signal. This result persisted across all blocks of test trials, a finding consistent with many demonstrations of the stimulus familiarization effect and with expectations from an orientation-reaction theoretical framework. These findings would seem to offer a reasonable and simple explanation of the stimulus familiarization effect: Repeated presentation of a stimulus diminishes its alerting capacity, as evidenced both by the slower responses it elicits as a target and by the slower responses that occur to other targets when it serves as a warning signal. Two difficulties with such an explanation emerge, however, when the findings are considered within a broader context. First, the stimulus familiarization effect itself is paradoxical because of

256

Daniel W Smothergill and Alan G. Kraut

evidence from a variety of studies on priming and “pathway activation” (Posner, 1978)that the encoding of a stimulus isfacilitated by its prior presentation. Facilitation as a normal effect of prior exposure seems not to fit with the alertness decrement effect demonstrated in Kraut’s second experiment. Second, although the altertness component of attention is assumed to be independent of the encoding component, the decrement in alertness brought about by familiarization is manifestly stimulus specific. If it were not, a familiarized stimulus would be just as effective a warning signal as a novel stimulus, and Kraut’s finding indicated that it is not. Posner (1978), too, has pointed out the importance of this aspect of Kraut’s results. Kraut proposed that the apparent paradoxical nature of the stimulus familiarization effect lay in the fact that two changes occur as a result of familiarization. One change is alertness decrement, which has already been seen to slow the response to a familiarized stimulus. The other change is encodingfacilitation, which was hypothesized to have a positive effect on response speed. The stimulus familiarization effect, then, according to this two-factor theory, is the net result of greater alertness decrement than encoding facilitation. Kraut’s third experiment was designed to demonstrate encoding facilitation within the traditional familiarization paradigm. The experiment was run as an exact replication of the stimulus familiarization effect, except for one difference. The difference was that each test trial began with the onset of an inverted T as a warning signal. The rationale was that with the alertness function controlled in this way, responses to the target stimuli would be relatively uninfluenced by their alerting capacities and would reflect, instead, speed of encoding. Kraut found that under these conditions, the familiarized stimulus was in fact responded to more rapidly than the novel stimulus. Thus, these experiments taken together showed that the stimulus familiarization effect is not the inevitable consequence of repeated stimulus presentation, but rather, the net result of changes brought about in two components of attention.

IV. Age, Stimulus Familiarization Effects, and Latent Inhibition The research described so far was conducted exclusively with children who were between 4 and 7 years old. An early study conducted in the Iowa laboratories failed to reveal an effect of stimulus repetition in a sample of female college students (Meyers & Joseph, 1968). Since the procedures in this study were modeled closely on those used in numerous successful child studies, this negative result implied the possibility of a developmental effect worth pursuing. On the other hand, it was also pointed out (Cantor, 1969; Lubow et al., 1975) that the number of familiarization trials used by Meyers and

Alertness and Encoding Effects in Children

251

Joseph was less than the minimum found needed to yield an effect with children. Slowed responding to familiarized stimuli in adults now has been shown a total of eight times in three independent studies (Cecil, Kraut, & Smothergill, 1984; Kraut & Smothergill, 1978; Kraut, Smothergill, & Farkas, 1981). In five of these cases (the 0 and 2500 millisecond conditions in Cecil et al.; Experiment 1 and the O-minute condition of Experiment 2 in Kraut & Smothergill; the color condition of Experiment 1 in Kraut et a/.), familiarized and novel colors were tested within a choice-reaction-time paradigm by means of procedures differing in no obvious respects from the original child studies of Cantor and Kraut. The remaining three cases involved simple reaction-time tests. Experiment 3 in Kraut et af. showed that familiarization resulted in slower responding to both colors and visually presented words when subjects responded to stimulus onset; Experiment 4 of that study showed the same effect for colors when subjects responded to stimulus offset. The consistency of these findings leaves little doubt that adults, too, show slowed responding to familiarized stimuli. Appearing to stand in contrast with this conclusion are findings from a number of studies of latent inhibition, which has been defined as “a decrement in learning performance which results from the nonreinforced preexposure of the to-be-conditioned stimulus” (Lubow, 1973, p. 398). Latent inhibition has been demonstrated in a variety of species, but it is difficult to obtain in adult humans. Lubow (1973) cited five attempts that failed, and suggested that special conditions are required during the preexposure phase to obtain the effect in adult humans. These conditions are described as ones that mask the to-be-conditioned stimulus (e.g., Schnur & Ksir, 1969). For example, Lubow, Caspy, and Schnur (1982) studied the effects of a masking procedure in which scrambled letters appeared between adjacent circles or squares during preexposure and had to be decoded to the word they formed. In a comparison condition, letters were absent. The circles and squares were composed of several dimensions that had different values during preexposure. Lubow et al. found that subjects who had decoded letters during preexposure required more trials to reach criterion on a task requiring discrimination of particular values on these dimensions. Other conditions in these experiments ruled out the possibility that decoding per se was responsible for this result. Thus, when to-be-discriminated stimuli were masked by a decoding task during preexposure, a latent inhibition effect was found. The subjects in this study were boys between 9.5 and 11 years old. Thus, masking appears necessary to produce latent inhibition in children at least as young as 9.5 years. Is masking always required to obtain latent inhibition in humans? Perhaps not, for Lubow et af. (1975) demonstrated that nonmasked preexposure had an inhibitory effect in first and second graders but not in adults. Unfortunately, the meaning of this finding is unclear for two reasons. First, Lubow

258

Daniel U! Smothergill and Alan G. Kraut

et al. used a criterion task of simple reaction time; the latent inhibition phenomenon, in contrast, is defined in terms of performance on associative learning (conditioning) tasks (see Lubow, 1973). Whether the mechanisms underlying conditioning and reaction time can be assumed to be sufficiently similar to generalize from one to the other is largely unknown. The fact that in adults, masking is required for latent inhibition but not for the stimulus familiarization effect would suggest that they are not. Second, the design of Lubow et al’s (1975) study actually precludes any conclusion regarding effects of preexposure on the spec@ preexposed stimulus. In their study, either a yellow or white light was presented 0, 3, 10, 20, or 30 times. The only target stimulus presented on the simple reactiontime task that followed was the particular light that had been preexposed. The fact that children’s performance was found to be inversely related to number of preexposure trials could either be an effect specific to the preexposed stimulus (the latent inhibition interpretation) or a nonspecific effect of the number of preexposure trials. In short, organizing the research findings by procedures rather than by study titles (familiarization or latent inhibition) reveals a substantial degree of consistency across age in findings. Reaction time to a repeatedly presented stimulus has been found to be slowed in studies in which (1) stimuli have been presented for mere observation 20 or more times prior to reaction-time testing, and (2) both the familiarized stimulus and a nonfamiliarized stimulus serve as targets. This finding has been obtained reliably in simple- and choice-reaction-time tasks, in both children and adults, without obvious masking procedures. Kraut’s (1976) results, along with the others discussed in the next section, strongly suggest that this inhibitory effect is due to a decrement in the alerting capacity of a familiarized stimulus. In contrast, slowed associative learning as a consequence of preexposure of task-relevant stimuli appears to have been demonstrated in humans only under conditions in which the to-be-discriminated stimuli were masked during preexposure. The youngest age at which children have been tested for latent inhibition appears to be 9.5 years; masking was required to produce the effect at that age. Until studies with younger children are done, we have no way of knowing if some threshold age exists, prior to which latent inhibition occurs without masking. Given that the masking requirement has no obvious theoretical basis (Lubow et al., 1982), the possible existence of an age threshold for the masking requirement also has no theoretical basis at the present time.

V. Mechanisms of Familiarization Effects As noted in Section 111, Kraut (1976) provided evidence for a two-factor theory of stimulus repetition effects. This is essentially a balance theory, in

Alertness and Encoding Effects in Children

259

that the outcome of any particular familiarization experiment is said to be the net result of opposing effects on alertness and encoding. If the effect on alertness is stronger, responses to the familiarized stimulus should be slowed; if the effect on encoding is stronger, responses to the familiarized stimulus should be facilitated. Conceivably, too, a null result could occur if the two effects were equivalent in magnitude. Put in this way, the obvious problem is that the theory can account for virtually any experimental outcome by assuming that one, the other, or neither effect was dominant. The strategy pursued to resolve this problem has been to make use of converging experimental operations designed to influence particular component processes. A . DELAYED TESTING

Kraut and Smothergill (1978) familiarized adults to color stimuli and tested choice-reaction time to familiarized and novel colors. Testing occurred either immediately after familiarization or 15 or 30 minutes later. Posner and Boies (1971) had suggested that alertness is particularly phasic; therefore, Kraut and Smothergill reasoned that the effect of familiarization on alertness might be expected to dissipate rather quickly, with the result that the encoding effect would have a greater influence on response when testing was delayed. Results supported this expectation. When testing occurred immediately, the familiarized stimulus was responded to more slowly than the novel stimulus. But at both delayed testings, the familiarized stimulus was responded to more quickly than the novel stimulus. B. NOMINAL AND FUNCTIONAL STIMULI

A second converging operation, based on the distinction between nominal and functional stimuli (Underwood, 1963), was designed to demonstrate that encoding is indeed the locus of the facilitation brought about by familiarization. One study (Smothergill & Kraut, 1981), was based on findings Lasky (1974) obtained with patterns he created by transforming a prototype. Lasky found that recognition memory for these patterns depended not on whether they had actually been presented during the acquisition phase but on their transformational distance from the prototype. The prototype, which had not been presented during acquisition, received the highest rating as having been seen before. Other patterns were recognized as seen before as an inverse function of the number of transformations by which they differed from the prototype. Thus, subjects encode prototypes as a consequence of experience with their transformations. In the Smothergill and Kraut study, adults were familiarized to several of Lasky’s transformation patterns and then performed a choice-reaction-time task requiring that target stimuli be judged as same (seen during familiarization)

260

Daniel U! Smothergill and Alan G. Kraut

or different (not seen in familiarization). Consistent with Lasky’s findings, these judgments were found to be based on the number of transformations by which a target differed from the prototype. This finding suggests that in a functional sense, the stimulus that was “familiarized” was the prototype. If the facilitating effect of familiarization is on encoding, it would then be expected that same responses to the prototype and patterns transformationally near it would be faster than dgjerent responses to patterns further removed from the prototype. Results confirmed this expectation. Moreover, same responses were not always faster than different responses. When only correct responses (i.e., those based on the nominal familiarized stimuli) were compared, the speeds of same and dijJ3rent responses were not different. Tho conceptually similar experiments, with words as stimuli, were reported by Kraut and Smothergill (1980). In one experiment, 7- and 8-year-olds and 11- and 12-year-olds were familiarized to five words belonging to either the category of body parts or that of animal names. A choice-reaction-time task followed, in which the target stimuli were a new word from the familiarized category and a word from the other category. Thus, the nominal target stimuli were both novel. A nondiscriminant warning signal was presented just prior to target onset as a means of controlling alertness. At both age levels, responses to the word from the familiarized category were found to be faster than those to the word from the nonfamiliarized category. A second experiment with children at the same age levels, with words drawn from the categories positive and negative evaluation, yielded the same result. In sum, three experiments with subjects ranging from 7 years to adult have shown the locus of the facilitation effect of familiarization to be encoding. Moreover, results from two delayed-testing conditions (Kraut & Smothergill, 1978) converge on the conclusion, drawn in Kraut’s original research, that the inhibitory effect of familiarization is an alertness phenomenon.

VI. Stimulus Characteristics In the research reviewed so far, a variety of experimental operations have been employed in testing the hypothesis of separate familiarization effects on alertness and encoding. Other experiments that have been performed indicate that the strengths of these effects vary as a function of stimulus characteristics. Smothergill and Kraut (1980) suggested that the relative strength of encoding facilitation might be greater for more salient stimulus dimensions, and the strength of alertness decrement might be greater for less salient dimensions. Kindergarten children were familiarized to a colored form, and then responded to four targets in a simple reaction-time paradigm. One of the target stimuli was the familiarized stimulus. The other three targets

Alertness and Encoding Eflects in Children

26 I

differed from the familiarized stimulus in (1) color but not form, (2) form but not color, or (3) both color and form. As expected, responses were slowest to the familiarized stimulus and fastest to the completely novel stimulus. Responses to the two partial change stimuli were found to be a function of stimulus salience, which had been assessed independently. To the degree that form was more salient than color, responses were faster in the condition in which form remained the same but color changed, and vice versa. Hence, these findings support the idea that relatively more encoding facilitation than alertness decrement accrues to a salient stimulus dimension as a result of repeated presentation. However, because the dimensions of only color and form have so far been tested, more than the usual caution is in order regarding the generality of this formulation. A study by Kraut ef al. (1981) provides further evidence that the strengths of familiarization effects vary as a function of stimuli. Separate groups of adults were familiarized to colors or visually presented words, and then were tested in a choice-reaction-time paradigm without a warning signal. As noted in Section V, responses for colors were found to be slower to familiarized than to novel stimuli. For words, however, responses were found to be faster to familiarized than novel stimuli. Two-factor theory would suggest that the encoding facilitation effect was stronger than alertness decrement for words. To test this interpretation, a second experiment was conducted in which familiarized and novel words served as warning signals in a choice-reactiontime task, in which the target stimuli were red and blue circles. The parlticular combination of warning signal and target appearing on any trial was unpredictable. If an alertness decrement effect occurs to familiarized words, then a familiarized word should be a less effective warning signal than a novel word. Results confirmed this expectation. Responses to target stimuli were slower with the familiarized word as warning signal. Two additional experiments by Kraut et al. extended the study of word familiarization effects to simple reaction-time paradigms. In one (Experiment 3), adults who were responding to stimulus onset were slower to react to familiarized words than to novel words-the opposite of what had been found for choice-reaction time. The investigators reasoned that the procedure of responding nondifferentially to stimulus onset may have mitigated against full encoding of the targets, and in turn favored the alertness decrement effect of word familiarization. Support for this possibility was obtained in a further simple reaction-time experiment in which subjects responded to the offset of familiarized and novel words. Under these conditions, no difference in response speed was obtained. In short, these studies with adults suggest that both alertness decrement and encoding facilitation occur when either words or colors are familiarized. However, the relative strengths of familiarization effects seem different for

Daniel W Smothergill and A/an G. Kraut

262

words and colors; encoding facilitation was found to be stronger for words, and alertness decrement was shown in earlier research to be stronger for colors. The significance of these findings lies in having extended the scope of twofactor theory beyond the domain of colored forms. How other types of stimuli are affected by repeated stimulus presentation, and whether two-factor theory might account for such effects, remain to be seen.

VII.

Familiarization as a Tool in Studying Reading Acquisition

The foregoing research has provided new information on the processes of alertness and encoding in both children and adults. We describe in this section some experiments that represent attempts to put that information to use to understand the nature of the psychological changes that occur when children learn to read. One of the key ideas in reading research since the mid-1970s directly implicates encoding processes as a locus of significant change. Known as automaticity theory (LaBerge & Samuels, 1974), this idea grants considerable importance to the observation that print is encoded in a slow, deliberate manner by novice readers, but quite rapidly and without apparent effort by skilled readers. According to automaticity theory, the difference is only incidentally one of speed or magnitude. More fundamentally, a qualitative change in encoding is said to occur. The claim is that beginning readers encode by means of conscious attention, but skilled readers encode unconsciously (automatically). Encoding automatically is thought to be advantageous in that conscious attention is freed for comprehension. The major evidence bearing on this hypothesis has come from interference studies in which subjects are instructed to ignore print and to respond instead to pictorial shape. For example, a line-drawn house within which the word CAT is embedded is presented and the task is to say quickly “house” rather than “cat.” The prediction derived from automaticity theory is that interference from print should be found in skilled readers because they will encode CAT automatically in spite of deliberate effort not to do so. Novice readers, in contrast, are expected to show little or no interference, to the extent that their encoding of print is deliberate. The rather surprising result from a number of studies is that evidence of interference from print appears quite early in reading acquisition, certainly by second grade (see Horn & Manis, 1987, for a review). Demonstrations that interference is absent at any level of reading skill are in fact rare (Smothergill, 1982).A number of explanations have been offered for these findings, ranging from the possibility that the encoding of print is to some degree automatic

Alertness and Encoding Effects in Children

263

from the very beginning of reading, to the possibility that novice readers, typically young, are unable to control attention in accordance with task demands (Reiner & Morrison, 1983). Whatever the explanation, findings from interference studies have provided little support for the hypothesis that word encoding changes qualitatively over the course of reading acquisition. As noted, some evidence in adults indicates encoding facilitation of familiarized words (Kraut et al., 1981), and in second- and sixth-grade children, a categorical facilitation effect of word familiarization is clearly localized in encoding (Kraut & Smothergill, 1980). Considered in light of theoretical interest in word encoding, these findings prompted a series of word familiarization experiments, comparing novice and more skilled readers (Kraut & Smothergill, 1986). First (6- and 7-year-olds) and fifth (10- and ll-year-olds) graders were compared in Experiment 1. The research was conducted in May, so the first graders had had about 1 academic year of formal reading instruction. All words used in this experiment, and in the ones that followed, were within the reading vocabularies of the first graders. Subjects were familiarized to a single word and tested on a choice-reaction-time task in which the familiarized word and a novel word were targets. No warning signal was given. The results were that fifth graders responded more rapidly to the familiarized word than to the novel word (a replication of the Kraut et al. finding in adults), but first graders responded more slowly to the familiarized word than to the novel word. Kraut et af. had shown that adults’ speeded responses to familiarized words concealed a n alertness decrement effect that was also present. This finding suggested that just the opposite might be the case in first graders; alertness decrement might be responsible for the inhibition of responses to familiarized words, and a weaker encoding facilitation effect might also be present. This possibility was tested by using the nondiscriminant warning-signal technique devised originally for Experiment 3 of Kraut’s 1976 study. First graders were familiarized to a single word and, again, tested on a choicereaction-time task in which the targets were the familiarized word and a novel word. Several hundred milliseconds prior to target onset, a red circle appeared as a warning signal. Kraut had found that under these conditions, young children’s responses were faster to a familiarized stimulus than to a novel stimulus, a reversal of the stimulus-familiarization effect. Kraut and Smothergill reasoned that a similar reversal might be expected here if first graders’ word encoding is facilitated by familiarization. Surprisingly, however, as in Experiment 1, familiarized words were responded to more slowly than novel words. Several possible reasons were considered for this result. The most plausible, from the perspective of two-factor theory, were those based on the assumption that the warning signal had simply been ineffective. Perhaps a different

264

Daniel W Smothergill and Alan G. Kraut

kind of warning signal, or a different interval between the warning signal and the target, would be necessary to show that the encoding of familiarized words really is facilitated in novice readers. Alternatives to two-factor theory were also considered; in particular, the intriguing suggestion from automaticity theory that word encoding changes qualitatively with reading acquisition. Given that the encoding of familiarized words is known to be facilitated in more advanced readers (Kraut & Smothergill, 1980; Kraut et al., 1981), could the slowed responding to familiarized words in novice readers represent an inhibition of encoding? As a test of this change-in-encoding hypothesis, first-grade readers were familiarized in Experiment 4 to a set of five words drawn from the category of either body parts or animal names. These were the same categories and words used by Kraut and Smothergill (1980). The children were then given a choice-reaction-time task in which both target stimuli were novel. One of the targets was a new word from the familiarized category; the other was a word from the nonfamiliarized category. The result was that responses to the word from the familiarized category were slower than those to the word from the nonfamiliarized category. Because neither target had been presented during familiarization, this effect must reflect the influence of familiarization on semantic encoding. The slowed responding of novice readers to familiarized words seems, therefore, not to be an effect of alertness decrement but, instead, the result of encoding inhibiton. In sum, these findings provide support for the hypothesis that word encoding changes qualitatively with reading skill. Although this conclusion is quite consistent with the central claim of automaticity theory, how the mechanisms of automaticity theory are represented in these findings is not at all clear. In fact, a considerable gulf appears to separate the two. Automaticity theory suggests that with increased reading skill, the dependence of word encoding on conscious attention gives way to automatic encoding. The familiarization findings are that word encoding is inhibited in novice readers but facilitated in more advanced readers (Kraut & Smothergill, 1980, 1986; Kraut el al., 1981). We see no obvious reason why familiarization should inhibit conscious-attention encoding but facilitate automatic encoding, and we leave this as a problem for others to ponder. In the meantime, an important point to emphasize is that the familiarization findings are consistent with the hypothesis of qualitative change in word encoding. Research aimed at explicating the reason for the shift from inhibition to facilitation of word encoding should be pursued. As a start in that direction, the curious breakdown of two-factor theory with respect to the critical results in beginning readers would seem a good place to begin.

Alertness and Encoding Eflects in Children

265

VIII. Summary and Conclusions The findings reviewed here suggest a broadened view of the stimulusfamiliarization effect as one of a general class of phenomena resulting from repetition-induced changes in the processes of alertness and encoding. Cantor’s original observation of slowed responding to familiarized stimuli can be attributed o n this account to experimental conditions under which the magnitude of alertness decrement is greater than encoding facilitation. Exactly the opposite result can be obtained by arranging conditions so as to highlight the encoding facilitation effect relative t o the alertness decrement effect. For example, faster responding to familiarized stimuli has been found under conditions in which (1) the imperative stimulus is preceded by a neutral warning signal or, (2) a delay is imposed between the familiarization and the test phases. Earlier, it was noted that the experimental technique of repeated stimulus presentation is widely used in the study of infant perception, with little concern for the mechanisms responsible for what is generically referred to as habituation. Posner and Rothbart (1980) made a similar observation and suggested, on the basis of Kraut’s (1976) results, that Habituation in the infant can be viewed as a reduction in the flow of information from the recognition pathway [encoding] into the alerting system. Input is still processed along the recognition pathway, but its failure to activate the alerting system reduces the availability of the central processor and hence of nonhabitual responses. (p. 28)

We have developed a model of attention that accounts for our findings and embodies Posner and Rothbart’s general idea. It suggests that both the alertness and the encoding functions must reach separate threshold levels for a correct recognition response. The time required to get to each threshold level is changed as a function of stimulus repetition. Repetition slows alertness “rise time,” but it speeds encoding “rise time.” In comparing the responses to a familiarized color versus those to a novel color, the model can be seen in Fig. la. Because any response must await an appropriate level of alertness, the familiarized stimulus is responded to more slowly than the novel one, even though encoding rise time to the familiarized stimulus is fast. Either a delay between familiarization and test or a warning signal acts to mitigate the effects of familiarization on alertness. Alertness rise time for the familiarized stimulus is similar to that for the novel stimulus, but encoding rise time for the familiarized stimulus is faster than for the novel. The result is that the familiarized color is responded to more quickly than the novel one (see Fig. lb). Our words-versus-colors manipulation tested the effect of a change in complexity of the repeated stimulus, and it is modeled in Fig. lc. We first assumed

266

Daniel W! Smothergill and Alan G. Kraui a

hreshuld lcvcl ncedcd or correct rc>pon\e

time

b

hreshuld level needed or cotrrcr msporisc

tiine

lirrahold level needed LITcorrcci respon\e

C

time

Alertness and Encoding Effects in Children

267

that words are more complex than colors, that complexity has more of an effect on encoding than on alertness, and that the effect is to generally slow encoding rise time. It takes longer, then, to encode words than colors, but the familiar word is encoded faster than the novel. Tho-factor theory has now been tested under a variety of converging operations, and it has fared reasonably well. The most important exception is seen in the case of novice readers familiarized to words. Here, the result is encoding inhibition rather than alertness decrement and encoding facilitation. The fact that advanced readers respond more quickly to familiarized words under the same conditions is consistent with the claim that word encoding changes qualitatively with reading skill. The nature of the mechanism underlying this change is, however, far from clear. Aside from the aforementioned findings with words, the evidence for any interesting developmental changes in the effects of repeated stimulus presentation is weak. Indeed, developmental research on the processes of alertness and encoding is itself quite sparse (Posner, 1978). Morrison (1982) reports evidence that alertness both peaks more slowly and declines more quickly in 5-year-olds than in adults. As he notes, earlier suggestions of developmental differences in encoding are potentially compromised by these findings because alertness was not controlled. Finally, although the affective aspects of familiarity have received only limited attention in the child development literature (e.g., Cantor, 1968), they have been a major interest in “mere exposure” studies in social psychology. Familiar stimuli are typically found to be rated as better liked in these studies. Kunst-Wilson and Zajonc (1980) reported that stimuli presented under conditions such that subjects were unable to recognize them as having been seen before were still rated as better liked than novel stimuli. They interpreted this as supporting a distinction between affective and cognitive processes. In discussing how this affective process might function, Zajonc proposed a mechanism that readers of this review will recognize as-well-familiar: “When we encounter something new we become alert. We tense up, an unpleasant feeling. When we encounter it again, our alertness diminishes and with it the feeling of unpleasantness. So we become more positive towards the object” (Hall, 1987, p. 86). That the same mechanism should be advanced, quite independently, in accounts of two heretofore separate but similar research literatures is probably more than coincidental. Both research traditions may well be on the right track, and greater efforts should be made to view each literature in light of Fig. I . 7iuo-factor model of sfimulus repetition: (a) effects for color stimuli; (b) mitigation of alertness decrement; (c) effects for word stimuli. Broken lines indicate alertness, and solid lines indicate encoding. NS, novel stimulus; FS, familiarized stimulus.

268

Daniel W Smothergill and Alan G. Kraut

the other. For example, are encoding facilitation effects detectable as a consequence of “mere exposure”? Also, how might researchers further investigate the suggestion that alertness decrement, observed by us and others in familiarization studies, is actually an affective phenomenon? Perhaps most tantalizing, could the distinction Zajonc has proposed between affective and cognitive processes be mirrored in the difference between the alertness and the encoding effects of familiarization?

ACKNOWLEDGMENTS Much of the authors’ research reported here was supported in part by National Institute of Child Health and Human Development Grant HD06142 and by two National Institutes of Health Biomedical Research Support Grants from Virginia Polytechnic Institute and State University. We are grateful to the children, teachers, and school administrators in Oswego and LaFayette, New York, and Giles County, Virginia, for cooperating in this research.

REFERENCES Bertenthal, B. I., Profitt, D. R., Spetner, N. B., & Thomas, M. A. (1985). The development of infant sensitivity to biomechanical motions. Child Development, 56, 531-543. Bogartz, R. S., & Witte, K. L. (1966). On the locus of the stimulus familiarization effect in young children. Journal of Experimental Child Psycholop, 4, 317-331. Broadbent, D. E. (1958). Brception and communication. Oxford: Pergamon. Cantor, G. N. (1963). Responses of infants and children to complex and novel stimulation. Advances in Child Development and Behavior, 1, 1-30. Cantor, G. N. (1968). Children’s “like-dislike” ratings of familiarized and nonfamiliarized stimuli. Journal of Experimental Child Psychology, 6, 651-657. Cantor, G. N. (1969). Effects of stimulus familiarization on child behavior. In J. P. Hill (Ed.), Minnesota Symposia on Child Psychology (Vol. 3, pp. 3-30). Minneapolis: University of Minnesota Press. Cantor, G. N., & Cantor, J. H. (1964). Effects of conditioned-stimulus familiarization on instrumental learning in children. Journal of Experimental Child Psychology, 1, 71-78. Cecil, L. S., Kraut, A. G.,& Smothergill, D. W. (1984). An alertnessdecrement hypothesis of response inhibition to repeatedly presented stimuli. American Journal of Psychology, 97, 391-398. Cohen, L. B., & S t r a w , M. S. (1979). Concept acquisition in the human infant. Child Deve[opment, 50, 419-424. Flavell, J . H. (1985). Cognitive development (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall. Hall, E. (1987). Growing and changing What the experts say. New York: Random House. Horn, C. C., & Manis, F. R. (1987). Development of automatic and speeded reading of printed words. Journal of Experimental Child Psychology, 44, 92-108. Jeffrey, W. E. (1968). The orienting reflex and attention in cognitive development. Psychological Review, 15, 323-334. Jeffrey, W. E. (1976). Habituation as a mechanism for perceptual development. In T. J. Tighe

Alertness and Encoding Effects in Children

269

& R. N. Leaton (Eds.), Habituation: Perspectivesfrom child development,animal behavior, and neurophysiology (pp. 279-296). Hillsdale, NJ: Erlbaum. Kellman, P. J., &Short, K. R. (1987). Development of three-dimensional form perception. Journal of Experimental Psychology: Human Perception and Performance, 13, 545-557. Kraut, A. G. (1976). Effects of familiarization on alertness and encoding in children. Developmental Psychology, 12, 491-496.

Kraut, A. G., & Smothergill, D. W. (1978). A two-factor theory of stimulus repetition effects. Journal of Experimental Psychology: Human Perception and Fkrformance, 4, 191-197. Kraut, A. G., & Smothergill, D. W. (1980). A new method for studying semantic encoding in children. Developmental Psychology, 16, 149-150. Kraut, A. G., & Smothergill, D. W. (1986). Qualitative change in word encoding with reading skill. Journal of Experimental Child Psychology, 24, 416-433. Kraut, A. G., Smothergill, D. W., & Farkas, M. S. (1981). Stimulus repetition effects on attention to words and colors. Journal of Experimental Psychology: Human Perception and Pep formance, 7, 1303-1311. Kunst-Wilson, W. R., & Zajonc, R. B. (1980). Affective discrimination of stimuli that can not be recognized. Science, 207, 465-469. LaBerge, D., & Samuels, S. J. (1974). Toward a theory of automatic information processing in reading. Cognitive Psychology, 6, 293-323. Lasky, R. E. (1974). The ability of six-year-olds and adults to abstract visual patterns. Child Development, 45, 626-632. Lubow, R. E. (1973). Latent inhibition. Psychological Bulletin, 79, 398-407. Lubow, R. E., Alek, M., & Arzy, J. (1975). Behavioral decrement following stimulus preexposure: Effects of number of preexposures, presence of a second stimulus, and interstimulus interval in children and adults. Journal of Experimental Psychology Animal Behavior hcesses, 104, 178-188. Lubow, R. E., Caspy, T., & Schnur, P. (1982). Latent inhibition and learned helplessnessin children: Similarities and differences. Journal of Experimental Child Psychology. 34, 231-256. Meyers, W. J., & Joseph, L. J. (1968). Response speed as related to CS-prefamiliarization and GSR-responsivity. Journal of Experimental Psychology, 78, 375-381. Miller, D. J. (1972). Visual habituation in the human infant. Childfkvelopment, 43, 483-493. Morrison, F. J. (1982). The development of alertness. Journal of Experimental Child Psychology, 34, 189-199. Posner, M. I. (1978). Chronometric explorations of mind. Hillsdale, NJ: Erlbaum. Posner, M. I., & Boies, S. (1971). Components of attention. Psychological Review. 78, 391-408. Posner, M. I., & Rothbart, M. (1980). The development of attentional mechanisms. In J. H. Flowers (Ed.), Nebraska S.vmposium on Motivation (pp. 1-51). Lincoln: University of Nebraska Press. Reiner, M. B., & Morrison, F. J. (1983). Is semantic interference really automatic? Bulletin of the Psychonomic Society, 21, 271-274. Ruff, H. (1986). Components of attention during infants’ manipulative exploration. Child Develop ment, 57, 105-114. Schnur, P., & Ksir, C. J. (1969). Latent inhibition in human eyelid conditioning. Journal of Experimental Psychology, 80, 388-389. Smothergill, D. W. (1982). Perception. In R. Vasta (Ed.), Straregies and techniques of childstudy (pp. 29-53). New York: Academic Press. Smothergill, D. W., & Kraut, A. G. (1980). Functional significance of dimensional dominance hierarchies. Merrill-&her Quarterly, 26, 197-204. Smothergill, D. W., & Kraut, A. G. (1981). Stimulus repetition and encoding facilitation: Locus of the effect. Canadian Journal of Psychology, 35, 93-98.

270

Daniel u! Smothergill and Alan G. Kraut

Thompson, R. F., & Spencer, W. A. (1966). Habituation: A model phenomenon for the study of neuronal substrates of behavior. Aychological Review, 13, 16-43. Underwood, B. J. (1963). Stimulus selection in verbal learning. In C. N. Cofer & B. S.Musgrave (Eds.), Wrbal behavior and learning: Problems and processes (pp. 33-48). New York: McGraw-Hill. Witte, K. L., &Cantor, G. N. (1967). Children’s response speeds to the offset of novel and familiar stimuli. Journal of Experimental Child Psychology, 5, 372-380.

IMITATION IN INFANCY A CRITICAL REVIEW

Claire L. hulson DEPARTMENT OF PSYCHOLOGY QUEENS COLLEGE AND THE GRADUATE SCHOOL CITY UNIVERSITY OF NEW YORK FLUSHING, NEW YORK 11367

Leila Regina de Paula Nunes DEPARTAMENTO DE PSICOLQGIA UNIVERSIDADE FEDERAL DE SAO CARLOS SAO CARLOS 13560, SAO PAULO, BRAZIL

Steven E Warren DEPARTMENT OF SPECIAL EDUCATlON GEORGEPEABODYCOLLEGE VANDERBILT UNIVERSITY NASHVILLE, TENNESSEE 37203

I. INTRODUCTION 11. EARLY CONCEPTUALIZATIONS OF INFANT IMITATION 111. DEVELOPMENTAL TRENDS IN DIVERSE IMITATIVE RESPONSES A. GESTURAL IMITATION: USE OF SUBJECTS OF DIFFERENT AGES B. VOCAL IMITATION C. DISCUSSION OF OVERALL DEVELOPMENTAL TRENDS IV. VARIABLES AFFECTING INFANT IMITATIVE PERFORMANCE A. INFANT’S ATTENTION TO THE MODEL B. LATENCY OF IMITATION C. INFANT’S DEVELOPMENT OF OBJECT PERMANENCE D. CONDITIONS O F AN INFANT’S UPBRINGING E. SUMMARY V. IMITATION AND MOTHER-INFANT INTERACTION VI. METHODS OF EVOKING IMITATION IN INFANTS VII. NEONATAL IMITATION O F NONVISIBLE ACTIONS VIII. PAST AND FUTURE TRENDS IN THE STUDY OF INFANT IMITATION REFERENCES 271 ADVANCES IN CHILD DEVELOPMENT AND BEHAVIOR, VOL. 22

Copyright 0 1989 by Academic Press, Inc. All rights of reproduction in any form reserved.

212

Claire L. fbulson et al

I. Introduction Imitation plays an important role in the development of early socialization and language. Imitative behavior in infants and young children has been extensively studied from different theoretical perspectives (Yando, Seitz, & Zigler, 1978). Cognitive researchers stress the operations of covert cognitive processes underlying imitation (Guillaume, 1926/1971; Piaget, 1946/1951), and behavioral researchers emphasize observable and measurable events in the demonstration of functional relations between imitative behavior and environmental stimuli. Little agreement has emerged on the criteria defining imitation. The one common element among definitions of imitation is that it involves similar responses from two organisms on the same occasion. This common element is insufficient if we are to view imitation as a tool in the acquisition of more adult-like social and linguistic skills in children. Imitation must be more than a coincidental emission of two similar responses by two individuals. Specifically, the behavior of the model must be shown to influence the behavior of the imitator. More specifically, the fine-grained topography of the model’s behavior must be shown to influence the infant’s behavior. If the model’s behavior functions merely to signal when, rather than how, to respond, we refer to the infant’s response as “matched-dependent behavior” (Miller & Dollard, 1941), rather than imitation. In this article, imitation is generally defined according to Uzgiris’s observation that “imitation is said to occur whenever a subject duplicates the behavior enacted by a model as a result of having observed the model” (1973, p. 599). This definition, although broad, does underscore the potential of imitation as a mechanism for the acquisition of new behavior. A more specific definition that focuses on the power of imitation as a learning mechanism is embodied in the term “generalized imitation,” as first proposed by Baer, Peterson, and Sherman (1967, p. 405) and later elaborated by Baer and Deguchi (1985). Generalized imitation occurs when (1) a potential imitator observes the responding displayed by a model, (2) an imitator’s responding temporally follows the demonstrated responding, (3) the topography of an imitator’s responding is functionally controlled by the topography of a model’s responding, (4) this topographical control is exhibited over a wide variety of modeled responses, and ( 5 ) imitative responses that have seldom or never been reinforced are performed as long as other imitative responses continue to be reinforced. The generalized imitation paradigm has been invoked in only a few infant studies. In the great majority of studies reviewed here, the authors use the term imitation roughly in the sense of the Uzgiris definition.

Infant Imitation

273

Studies of imitation have encompassed diverse populations, such as normal preschool and school-age children, exceptional children, and normally developing infants. Studies conducted with normal preschool and school-age children have shown that imitative performance is affected by several factors, including (1) contingent reinforcement delivered to the child (Baer & Sherman, 1964); (2) social setting events (Steinman, 1970); (3) contingent reinforcement delivered by the model and observed by the child (Bandura, 1965; Egel, Richman, & Koegel, 1981); (4) previous imitation of the child by the model (Hallahan, Kauffman, Kneedler, Snell, & Richards, 1977; Parton & Priefert, 1975); (5) the child’s previous positive interaction with the model (Bandura & Huston, 1961); (6) the child‘s identification with the model (Kornhaber & Schroeder, 1975); (7) the power or dominance of the model (Hetherington, 1975); (8) the child’s verbalization about the model’s responses during observation of the model (Coates & Hartup, 1969); and, fundamentally, (9) similarity between the model’s and the child’s behavior (Baer & Deguchi, 1985). Even though the foregoing studies of normally developing children have provided important data for identifying variables that affect imitation, they do not demonstrate how the process of learning to imitate occurs, as their subjects may be presumed to have already displayed imitative skills prior to their participation in the studies. In an alternative method of investigating the acquisition of imitation, many researchers have conducted studies with exceptional, nonimitating subjects. Most of these studies have been conceptualized within the operant learning paradigm and have involved single-subject research methodology. In these studies, researchers have established imitation in nonimitating autistic and schizophrenic children (Hewett, 1965; Lovaas, 1977; Lovaas, Berberich, Perloff, & Schaeffer, 1966; Lovaas, Freitas, Nelson, & Whalen, 1967), and mentally retarded individuals (Baer et al., 1967; Garcia, Baer, & Firestone, 1971; Martin, 1971; Nelson, Cone, & Hanson, 1973; Striefel & Phelan, 1972) through training procedures such as positive reinforcement, modeling, shaping, fading (Terrace, 1963), and putting the child through the motions (Konorski & Miller, 1937). Functional variables identified in studies of severely handicapped children may or may not have any role in the acquisition of imitation in normally developing children. Nevertheless, even if the learning processes are similar in exceptional and normally developing children, the handicapped children in the preceding studies were at least 4 years of age. An experimenter cannot realistically observe a child for 4 years to document that he or she does not already have an imitative repertoire. An alternative way to approach the study of the acquisition of imitation is to study infants (Parton, 1976). In fact, Hartup and Coates (1970) maintained that we must study infants directly if we are to understand the origins of imitation in child development.

214

Claire L. h h o n ef al.

The purpose of this article is to investigate what is known about the origins of imitation in infancy. Accordingly, a review of the literature was conducted to examine the following: (1) early conceptualizations of infant imitation, (2) the developmental course of diverse imitative responses, (3) variables affecting infant imitative performance, (4) imitation and mother-infant interaction, ( 5 ) procedures that evoke imitative responding in infants, (6) the issue of neonatal imitation of nonvisible actions, and (7) past and future trends in the study of infant imitation.

11. Early Conceptualizations of Infant Imitation Infant imitation has been the subject of descriptive research since the end of the past century (Hartup & Coates, 1970). Guillaume (1926/1971), Piaget (1946/1951), and Valentine (1930), within a cognitive perspective, have made important contributions toward understanding the origins of imitation in infancy. These authors gathered data through longitudinal and naturalistic observations of their own children, coupled with ad hoc experiments. Imitation, in Piaget’s view, represents an early manifestation of intelligence. Although Guillaume and Piaget both stressed the close relationship between imitation and intelligence, they had different views of the continuity versus discontinuity issue in the development of imitation. For Guillaume, an infant’s early imitation is merely a continuation of “circular reactions.” Self-imitation is not directly related to later imitation, or “true imitation,” which is the intentional reproduction of a model (1926/1971, pp. 77, 107). In contrast, Piaget defended the thesis of continuity, stating that the function of imitation-as part of the accommodation process-is the same in all stages, although it is less differentiated from the assimilation process during the first three stages of the sensorimotor period (1946/1951, p. 51). Guillaume, Piaget, and Valentine agreed that during the first few months, an infant imitates models if these models can be accommodated to a schema that the infant has already formed. They held that the chances of success in early vocal imitation increased if a familiar sound was produced by the model immediately after the infant emitted it spontaneously. Also, actions that yield clear perceptual feedback were believed more likely to be imitated than those that d o not. Guillaume and Piaget believed that during the first months of life, spontaneous imitation of the movements of others does not occur if the infant cannot see himself or herself make them, as in facial gestures. Training, however, might help an infant, in the earlier months, to imitate these nonvisible movements. Piaget referred to those imitative responses that are established through training as “pseudo-imitations.” Such trained responses would involve no direct assimilation of a model to an infant’s own activity,

Infant Imitation

275

and the responses would occur only under the control of “continual stimulation” (1946A951, p. 18). The ability to imitate facial gestures without training should appear during Piaget’s fourth stage of infant intelligence (8-12 months) because of the reciprocal assimilation of visual data to tactile-kinesthetic data. This process would allow the child to grasp the relationship between what he or she sees in someone else’s face and the motor impression of his or her own face in making a similar gesture. In addition, during the fourth stage, a greater number of schemas and their coordination, along with an increasing tendency to investigate objects, would enable the infant to accommodate these schemas to partially new models. The imitation of new models-that is, those actions that an infant has never emitted spontaneously-would begin during Piaget’s fifth stage (12-18 months). In this stage, an infant would advance beyond simple application of existing schemas through controlled and systematic trial and error, to true imitation. Similarly, purposive or true imitation, was defined by Guillaume (1926A971, p. 107) and Valentine (1930, p. 107) as occurring when a child’s responses are no longer subordinated to their results-or consequences-but to acts performed by a model. Finally, in the sixth stage, deferred imitation would appear-that is, an infant would be able to reproduce a model when it is no longer present. According to Piaget, a model perceived externally would be replaced by an internal model-the mental image. In fact, a child’s first mental images were held to be the internalization of a model. The studies of Guillaume, Piaget, and Valentine formed the basis for many of the more recent investigations of infant imitation. In fact, in approximately 30 of the 65 studies reviewed for this article, Piagetian theory was used as the main reference point. Only four studies (Hursh & Sherman, 1973; Nunes, Poulson, Nunes, Almeida, & Warren, 1985; Poulson & Kymissis, 1988; Waxler & Yarrow, 1975) were based on behavioral learning theory. The two major categories of imitative responding found in the literature were gestural and vocal. A summary of major investigative procedures, their findings for gestural and vocal imitative responding, and a discussion of overall developmental trends is presented in the following section.

111. Developmental Trends in Diverse Imitative Responses A. GESTURAL IMITATION: USE O F SUBJECTS O F DIFFERENT AGES

Change in gestural imitative performance related to age was the major focus of 15 studies. Three of these age-related studies were longitudinal, nine were

216

Claire L. Ibulson et al.

cross-sectional, and three (Abravanel, Levan-Goldschmidt, & Stevenson, 1976; Field, Goldstein, Vega-Lahr, & Porter, 1986; McCall, Parke & Kavanaugh, 1977) involved both types of designs. Although most of the investigators found that the frequency and accuracy of imitative responding increased with age, the precise pattern and level of performance differed as a function of the type of response modeled. Within the domain of gestural imitation, different developmental trends were found when the model involved: (1) the use of objects, (2) diverse parts of the body, (3) combined actions, (4) sound or form products, ( 5 ) familiar actions, (6) appropriate actions with objects, (7) use of different agents, and (8) deferred imitation. A description of the findings for each gestural response type follows.

1. Use of Objects in Gestural Imitation McCall et al. (1977) found that imitation of simple gestures with objects increased both in frequency and accuracy in a linear trend from 12 to 24 months in longitudinal and cross-sectional samples. Abravanel et al. (1976), using these two kinds of samples, reached the same conclusion, adding that this upward trend in accuracy begins at 6 months of age.

2. Use of Diverse k r t s of the Body Studies on imitation of body gestures without objects indicated different acquisition trends, according to the part of the body involved. Four studiesthree longitudinal (Field et al., 1986; Jacobson, 1979; Maratos, 1973) and one cross-sectional (Fontaine, 1984)-conducted with infants under 6 months of age, revealed consistently that the imitation of mouth movements (tongue protrusion, mouth opening, lip widening, and pouting) declined with age. In contrast, in a study by Fontaine (1984), the imitation of cheek swelling and eye closing, already present in 2-month-olds, was maintained until 5 months of age. The imitation of hand movements was investigated in two studies, one with infants whose ages varied from 1; to 3; months (Jacobson, 1979) and another with infants of 1 to 6 months (Fontaine, 1984). Jacobson found an increase across groups in rate per minute of imitation of manual gestures, but Fontaine did not. Facial gestures (tongue protrusion, mouth opening, and eye blinking) and hand gestures (hand opening and chest and chin tapping) were modeled for infants between the ages of 1 and 5 months in two studies conducted by Abravanel and Sigafoos (1984). Regardless of the number of repetitions of each model, the authors found that only tongue protrusion was imitated, and that only in the 1- to 1.5-month-old sample. The imitation of facial and manual actions was also investigated with older infants in the Abravanel et al. (1976) study of 6- to 15-month-olds. Movements

Infant Imitation

211

of the lips, head, and eyes were imitated significantly more accurately by the older infants, but imitation of mouth, tongue, and hand movements did not increase in accuracy over time. In contrast, imitation of modeled actions without objects, when compared with imitation of gestures with objects, showed a very slight increase in accuracy from 6 to 20 months (Abravanel et al., 1976; Rodgon & Kurdek, 1977). The relatively small progress made in the imitation of body gestures without objects may be due to an overrepresentation of nonvisible movements among the latter. Most of the body gestures without objects involved movements of the mouth, head, lips, and tongue, which the infant could not see himself or herself performing. Tongue protrusion and mouth opening also were modeled by mothers of infants less than a month old (14-21 days) in a study by Heimann and Schaller (1985). The authors obtained no evidence of differential imitation by the infants as a group, but they obtained imitation by individual infants within the group. The Heimann and Schaller study underscores the point that large individual differences are found among infants and indicates a need for further individual-subject analyses of imitation.

3. Use of Combined Actions Imitation of combined (two-part) actions develops more slowly across age than imitation of simple actions, and it shows its greatest increases both in frequency and accuracy between 15 and 18 months, according to McCall et al. (1977). Fenson and Ramsay (1981), in studying 15- and 19-month-olds, and McCabe & Uzgiris (1983), testing 12-, 17-, and 22-month-old infants, confirmed the trend described by McCall et al. (1977). In a longitudinal investigation, Uzgiris (1973) found that at 14 months of age, all infants in her sample imitated combined actions, which they had never displayed spontaneously before.

4.

Use of Responses with Sound or Form Products

McCall et al. (1977) presented models who used both objects that produced sounds (small cans) and objects that produced forms (puzzles). No differences in frequency or accuracy of imitation were found within or between age groups. In comparing imitative performance of pairs of actions producing or not producing sounds, Abravanel et al. (1976) obtained no clear evidence that the presence of sound increased the accuracy of imitative responding.

5. Use of Familiar Actions The dimension of infant familiarity with the modeled actions was examined in three studies. Infant familiarity with models refers to the spontaneous

Claire L. fbulson et al.

278

display of an action by an infant in the absence of modeling. Largo and Howard (1979) studied the imitation of functional and representational play by modeling single actions to 9- to 18-month-old infants. Regardless of age, the infants did not imitate either functional or representational actions unless they had already displayed them spontaneously. Fenson and Ramsay (1981) observed 12- to 15- and 19-month-olds and found that only the oldest infants imitated coordinated sequences they had never displayed spontaneously. Their finding contrasted with the results of Uzgiris’s (1973) study. Uzgiris reported that simple, familiar actions were imitated by all infants in her sample by 11; months, and that at 14 months, these subjects were able to imitate complex actions (i,e., two familiar actions coordinated in a sequence never spontaneously displayed before). Unfamiliar visible actions and unfamiliar invisible actions were imitated by all infants at 19 and 23 months, respectively.

6. Use of Appropriate Actions with Objects Two cross-sectional studies focused on the ways infants of different ages respond to models using socially appropriate and socially inappropriate actions with objects. Killen and Uzgiris (1981) presented simple actions of both kinds to infants 7;, 10, 16, and 22 months of age. A socially appropriate action with an object was, for example, pushing a car; drinking from a car was an inappropriate action. The investigations found a significantly increasing linear trend in accuracy of imitation with both types of models across age, although the models using appropriate actions were imitated earlier than the models with inappropriate actions. Bates, Bretherton, Shore, Snyder, and Volterra (1980) presented the Killen-Uzgiris task to a sample of 15-month-olds and found that appropriate gestures with objects were more frequently imitated than inappropriate gestures with objects. A replication of the Killen and Uzgiris (1981) study conducted by McCabe and Uzgiris (1983) failed to demonstrate an age trend in accuracy for both kinds of models with 12-, 17-, and 22-month-old subjects. Nonetheless, McCabe and Uzgiris replicated the Killen and Uzgiris finding that appropriate and inappropriate actions with objects were matched equally well by 22-month-old children.

7.

Use of Different Agents

Watson and Fischer (1977) examined the imitation of pretend acts with three different agents. During the modeling phase, the subjects, whose ages ranged from 14 to 24 months, were presented three agents-the experimenter, a doll, and a block-involved in three familiar actions-eating, sleeping, and washing. Four kinds of agents were identified in pretend play during a free-play period that followed the modeling phase: (1) self (action performed by the infant);

Infant Imitation

219

(2) active other (action performed by a doll); (3) passive other (action performed by the child with a doll); and (4) passive substitute (action performed by the child with a block). The duration of the pretend acts during free play showed that the use of self or passive other as agents decreased across age, and the use of passive substitute and active other as agents increased with age.

8. Deferred Imitation Deferred imitation of gestural models was evaluated by McCall et al. (1977). Following a modeling-with-toys phase, the experimenter presented the same toys, but without presenting the modeled responses. Then the infants were again exposed to the same toys while playing with their parents. In both situations, infants did not consistently exhibit deferred imitation until 24 months of age. Later, using (1) fewer toys, (2) formal experimental control groups with no modeling, and (3) identical procedures during pre- and posttesting, Meltzoff reported evidence of deferred imitation in 14-month-olds over a 24-hour delay (1985) and over a 1-week delay (1988b). Abravanel and Gingold (1985) produced evidence of deferred imitation in 12- and 18-month-olds over a 10-minute delay, and Meltzoff (1988~)reported deferred imitation in 9-month-olds following a 24-hour delay. Meltzoff (1988a) also demonstrated 24-hour delayed imitation in 12- and 24-month-old infants following televised modeling. In most cases, the authors discussed their findings with respect to cognitive capacities of infants, with a particular focus on memory. B. VOCAL IMITATION

The development of vocal imitation of voice pitch, cooing, babbling, novel sounds, and familiar and new words has been examined in five studies. Kessen, Levine, and Wendrich (1979) investigated the imitation of voice pitch in 23 infants ranging in age from 3 to 6 months. The subjects matched the target pitches more often than the nontarget pitches, with no age differences in pitch matching. In a longitudinal study, Maratos (1973) found an increasing trend in the percentage of imitation of another infant’s recorded babbling of “ma” from 1; to 54 months of age. Uzgiris (1973) reported that by 3 months, her infant subjects imitated cooing sounds, but they did not match babbling sounds until 13 months. Indeed, the McCall et al. (1977) study showed that vocal imitation progressed at a slower rate both in frequency and accuracy than gestural imitation between the ages of 12 to 24 months, with major increases in sound imitation occurring between 18 and 24 months. Similarly, in a cross-sectional study of infants 8, 14, and 20 months of age, Rodgon and Kurdek (1977) found more gestural than vocal imitation at each age group. Furthermore, they found significant differences in the frequency of imitation of repeated sounds and words by infants between 14 and 20 months of age, with sounds imitated earlier and more frequently than words. Imitation of

Claire L. hulson et al.

280

words by the 20-month-olds occurred at the infants’ current level of spontaneous production observed during a free-play situation in the Rodgon and Kurdek study. In a longitudinal study reported by Uzgiris (1973), all 19-montholds imitated novel sounds, all 23-month-olds imitated familiar words (words in the infants’ repertoires), and 9 of 12 24-month-olds imitated new words. C.

DISCUSSION OF OVERALL DEVELOPMENTAL TRENDS

Most of the developmental studies of infant motor and vocal imitation have shown that imitative performance increases in frequency and accuracy over time as a function of an infant’s current repertoire of responses, the types of responses modeled, and the social attributes of the objects used in modeling. Although only two studies (Abravanel et al., 1976; Watson & Fischer, 1977) actually included assessment of cognitive development (the object permanence concept), most of the authors inferred that changes in imitative performance are related to changes in cognitive structures. For example, McCall et al. (1977) suggested that the delay in imitating combined actions results from a delay in the cognitive linking of the two elements, rather than from any delay in motor development. In other words, the infant would be said to have not yet established “entityentity relationships.” Deferred imitation by 24-month-olds was explained in terms of the infants’ capacity to represent a model internally, and thus to transport it across time and across situations. The main threat to internal validity in these investigations was a high probability of the infants’ spontaneous emission of the target behavior independently of the model. If the modeling event functioned only as a cue (or discriminative stimulus) for when to respond, rather than how to respond, the infant’s response should not be called imitation, but only “matcheddependent behavior,” as described by Miller and Dollard (1941). The topography of the model’s behavior must be demonstrated to control not only the timing of the infant’s response, but also the topography of that response. To control for the possibility that infants might exhibit matched-dependent rather than imitative responding, four basic procedures have been used: (1) observation of a control group, (2) the inclusion of a baseline control procedure, (3) alteration of the modeling procedure, and (4) measurement of differential frequencies or proportions of individual responses. Specifically: 1. A control group is observed to determine whether it spontaneously emits the target responses modeled to the experimental group, as in the Abravanel et al. (1976) study. Of course, the problem of large individual differences among infants is ignored in this procedure. Nevertheless, it is better than no control at all. 2. Individual infants are observed in a free-play situation, prior to the modeling phase, to determine whether they spontaneously display the target

Infant Imitation

281

responses. This baseline control procedure has been used in most of the studies. The individual infant serves as his or her own control in this procedure. McCall et al. (1977) responded to the resulting information by presenting to the infant an alternative model if that infant spontaneously emitted the target response during free play. 3. McCall et al. (1977) extended the logic of the first two control procedures even further by using a modeling procedure in which, if an infant matched the target model, the experimenter modeled a slightly different response, and then repeated the target model, all in the same trial. If an infant matched all three responses, imitation was confirmed. 4. Abravanel and Sigafoos (1984), Fontaine (1984), and Field et af. (1986) measured the frequency of trials during which a specific facial expression occurred while other actions were modeled. A potential problem with this type of measurement procedure was borne out in their findings. Field, Woodson, Greenberg, and Cohen (1982); Field et af. (1986); and Fontaine (1984) reported that infants produced a targeted response during modeling of that response more often than other nontargeted responses, regardless of which infant response was targeted. Nevertheless, the sum of the combined nontargeted responses almost always was greater than the sum of the targeted responses. That is, the measure of infant responses that matched the model was generally exceeded by the combined measures of infant responses that did not match the model. If matching criteria were used, as has been the case in studies with older infants (e.g., Nunes et af., 1985; Poulson & Kymissis, 1988), these younger infants would not have been said to have shown matching. The effect of the model on the behavior of these younger infants, then, is very slight in comparison to the effect of a model on the matching behavior of the older infants. Of course, the age of the infant is only one possible factor contributing to differences between the two sets of studies.

IV. Variables Affecting Infant Imitative Performance Studies of infant imitation have shown that the following variables are associated with imitative performance: (1) the infant’s attention to the model, (2) the infant’s latency of imitation, (3) the infant’s development of object permanence, and (4) conditions of the infant’s upbringing. A. INFANT’S ATTENTION TO THE MODEL

Abravanel et al. (1976) found a positive correlation between imitation level and attention measures. Infants, regardless of age, paid more attention to modeled actions using objects, especially if the object produced sound, although sound was not consistently related to imitation. McCabe and Uzgiris

282

Claire L. &ulson et al.

(1983) also found that in all age groups (12, 17 and 27 months) and within all types of modeled gestures, the highest imitation scores were associated with high attention scores. B. LATENCY OF IMITATION

According to Abravanel et al. (1976), the latency of imitative responding following presentation of the model is an important criterion in determining true imitation, as opposed to matched-dependent behavior. These investigators included control groups who played with the same toys used in modeling for the experimental groups, but for whom no modeling was performed. Matched responses to a model were found to have significantly shorter latencies than similar responses not being modeled, which the investigators interpreted to indicate true imitation. Furthermore, the study showed an inverse relation between performance level and latency of response, as well as performance level and number of repetitions of the model required to produce the response. Thus, the higher the performance levels, the shorter the latencies and the fewer the repetitions of the model. C. INFANT’S DEVELOPMENT OF OBJECT PERMANENCE

The relationship between the development of object permanence and imitation was investigated in two studies. Abravanel et at. (1976) found a significant positive correlation between the mean scores of gestural imitation and the level of object concept, measured according to both Piaget’s (1946A951) and Gratch and Landers’s (1971) criteria. Nevertheless, when age was partialed out, the correlation was not significant. The relation between stages of object permanence assessed by the UzgirisHunt Scale and the steps of agent use (self, passive other, passive substitute, and active other) was studied by Watson and Fischer (197’7). They reported a moderate correlation (r = .62), but no “precise synchrony.” Age was also positively correlated with both measures. D. CONDITIONS OF AN INFANT’S UPBRINGING

Different conditions of a child’s rearing seem to be associated with the age at which infants display certain kinds of imitative performances. Paraskevopoulos and Hunt (1971) compared the influence of three conditions of child-rearing on a child’s imitative performance: the home, an orphanage with an infant-caretaker ratio of 10 : 1, and an orphanage with an infant-caretaker ratio of 3 : 1. All infants came from low socioeconomic backgrounds. The study revealed that different conditions of child rearing did affect the average

Infant Imitation

283

age at which infants achieved the successive levels of object construction and vocal imitation as shown with the Uzgiris-Hunt Scale. These conditions, however, were not an influential factor in the development of gestural, as opposed to vocal, imitation. Home-reared infants achieved the various levels of vocal imitation significantly earlier than orphanage-reared children. Infants from the low-ratio orphanage reached these levels earlier than those reared in the high-ratio orphanage. The psychological development of infants from middle-class and disadvantaged families was investigated by Wachs, Uzgiris, and Hunt (1971). The infants were 7, 11, 15, 18, and 22 months old; they were tested with the Uzgiris-Hunt Scale. Home conditions were assessed with a Home Stimulation Scale elaborated by Wachs et af. from the Caldwell Inventory of Home Stimulation. The difference in vocal imitation between the middle-class and the disadvantaged infants was significant at 15, 18, and 22 months of age and favored the middle-class infants. The authors related three home conditions to the development of infants: intensity of stimulation, variety of change in home circumstances, and opportunity to hear vocal signs for specific objects, actions, and relationships. This finding was supported by Tulkin and Kagan (1972), who showed that the lower the socioeconomic status of the family, the less frequently the mother vocalized to the infant. Thus, low frequency of caretaker-infant vocal interaction may explain the slower development of vocal imitation by infants reared in crowded orphanages, compared to the imitation scores of infants reared in a low-ratio orphanage where the caregiver probably had more time available to talk to the infants. E. SUMMARY

In summary, Abravanel et af. (1976) reported positive correlations between measures of infant attending and an infant’s imitation of a model, which appears to be enhanced by the use of objects in modeling. Furthermore, the highest levels of imitativeness were correlated with the shortest latencies and the fewest repetitions of the model. This suggests that if infants are attending, they respond quickly to a given model. It is less clear whether the concept of object permanence plays a strong role in infant imitation, based on the data available. Characteristics of the infant’s home environment also have been correlated with infant vocal imitation as measured by the Uzgiris-Hunt Scale. Data from Paraskevopoulos and Hunt (1971) indicate that the conditions of an infant’s upbringing are relevant, because home-reared infants or those in orphanages with a low infanthtaff ratio achieved vocal imitative milestones earlier than those in orphanages with a high infanthtaff ratio. Furthermore, Wachs,

284

Claire L. Bulson et al.

Uzgiris, and Hunt (1971) report that middle-class infants achieved milestones in vocal imitation earlier than infants from disadvantaged families. This observation is also supported by research by lhlkin and Kagan (1972), showing that the lower the socioeconomic status of the family, the less frequently the mother vocalized to the infant.

V. Imitation and Mother-Infant Interaction Only two sets of investigators (Moran, Krupka, Tutton, & Symons, 1987; Pawlby, 1977) examined imitation in the context of the mother-infant interaction literature (Newson, 1974). Pawlby observed in a longitudinal-descriptive study the frequencies and types of imitative sequences that occurred between mother and infant in a free-play situation. Subjects were seen once a week between the ages of 4 and 10 months. Imitative sequences were found to occupy only 16% of the interaction time. The number of infant-initiated, mother-imitated sequences was greater than the number of mother-initiated, infant-imitated sequences. The most frequent imitative sequence across the entire study consisted of speech sounds. Imitative sequences of facial movements occurred more frequently when the infant was between 4 and 6 months old, and such sequences involving hand movements and nonspeech sounds occurred more frequently when the infant was between 6 and 8 months old. Sequences involving manipulation of toys were more frequent when the infant was 8 and 10 months old. Moran et al. (1987) in a descriptive study of 13- to 16-month-old infants obtained 3-minute samples of free play between 20 mothers and their infants. Sequential probability analysis of broad categories of mother and infant behavior (mouth openings, lip movements, tongue protrusions, smiling, exaggerated eye movements, and vocalizations) revealed that infants produced smiling and exaggerated eye movements with greater probability when their mothers were already engaged in similar activity. There was little overlap in vocal activity between mother and infant. As in the Pawlby study, there was more evidence of mothers imitating infants than of infants imitating mothers. In both of the aforementioned studies of mother-infant interaction, observational, rather than experimental, methods were used, so that the causes of the matching patterns between mother and infant cannot be determined easily. For example, in the Moran etal. (1987) study, the mother might have anticipated infant smiling and responded by smiling herself, such that it appeared that the infant was imitating the parent. This possibility was suggested by the authors, who argued that an experimental analysis might be required to determine whether the mothers caused smiling in their infants through modeling procedures. Because much of the mother-infant interaction

Infant Imitation

285

literature is observational, rather than experimental, that literature might not be expected to yield much information concerning the causes of infant imitation. Nevertheless, reliable patterns of similar behavior exhibited by parents and infants may help alert experimental researchers to the most likely interactive activities upon which successful experiments might center. For example, the high-probability sequences involving smiling and exaggerated eye movements in the Moran et d.(1987) study suggest that those activities are ripe for experimental analysis.

VI. Methods of Evoking Imitation in Infants Most of the studies presented so far have involved basically the same method to evoke imitative responding from infants. Different models were demonstrated a few times to infants of different ages, whose responses are scored as exact imitation, partial imitation, or no imitation. Four studies, however, involved a different approach to the assessment and interpretation of changes in imitation: The procedures were designed to maximize the frequency of imitative responding of infants. Kaye and Marcus (1978) evoked imitation of mouth movements in 6-month-old infants, both by allowing the infants to control the timing of the presentation of the model and by repeating the model over many trials. The investigators observed the following behavioral sequence: (1) an orientation toward the experimenter, (2) a series of imitations of single features of the model, (3) a series of imitations of a string of two or more features of the model, and (4) integration of features and exact display of imitation. This sequence of accommodation across trials resembled the four-step course in imitative responding to models across age reported by Uzgiris (1973). In a second study, Kaye and Marcus (1981) presented motor and vocal models once a month t o infants between the ages of 6 and 12 months, in the same fashion as in their 1978 study. They again demonstrated that infants gradually acquired the full imitative model in a consistent order over trials and over months. This observation is consistent with the Piagetian theory that accommodation to models proceeds by a highly selective assimilation of their features, with infants defining for themselves what matches what. Of course, it is also consistent with the notion of reinforcement for successive approximations of a criterion-level response, although any reinforcement in the present studies would have been unplanned. In a series of studies (1982, 1983, 1986), Field and her colleagues have attempted to evoke imitation using procedures similar to those of Kaye and Marcus (1981). In the first two studies, infants were between 35 and 42 hours of age, and in the last study, they were between 2 and 6 months of age. With

286

Claire L. hulson et a/.

the neonates, the experimenter presented deep-knee bends and tongue clicks before each modeling trial. With the older infants, the experimenter simply presented the tongue-clicks. In all three studies, however, the modeling trial began with the experimenter fixing one of three facial expressions on her face until the infant looked away. Using a visual habituation procedure, Field and her colleagues obtained a higher than chance correspondence between infant and adult mouth movements, but this correspondence decreased from 2 to 3 months and again from 4 to 6 months, a finding consistent with the findings of Abravanel and Sigafoos (1984) and Fontaine (1984), as noted earlier in this article. These decreases in imitative behavior during early infancy have been discussed as possible artifacts of a procedure in which the developing infant becomes increasingly unattentive to an artificially fixed adult face, the infant develops facial expression blends, or the infant’s responding changes over time from early fixed-action-pattern (reflexive) behavior to a later form of socially interactive imitation (Field et al., 1986). The reasons for the apparent decline in early imitativeness remain a matter of conjecture. Although Field et al. (1983) found no evidence of differential reinforcement provided by adult female models for imitative responding by infants, uncontrolled contingencies of reinforcement could have had some influence on infant imitative behavior. Four studies focused directly on the relationship between imitation and reinforcement. In a within-group study, Waxler and Yarrow (1975) examined this relationship in the context of interactions of 19-month-old infants with their mothers. The subjects experienced two conditions: a play condition in which the mothers were asked to get their infants interested in certain toys and to emit specific responses with the toys; and a modeling condition, in which the mothers were asked to encourage their children to imitate them. Reinforcement was defined as positive verbal comments, smiles, and physical contact, independently from any contingency on behavior. During the play condition, group analysis revealed a strong association between frequency of reinforcement and frequency of imitation as well as between frequency of reinforcement and number of different kinds of modeled acts imitated. During the modeling condition, the same relation occurred, but it was not as strong. In addition, four patterns were found in the individual imitation learning curves: (1) the frequency of imitation increased after reinforcement for 37% of the subjects, (2) imitation increased in frequency with no reinforcement for 17%, (3) the frequency of imitation remained at very low levels whether with or without reinforcement for ll%, and (4) a reciprocal influence of reinforcement and imitation occurred in 34%. Besides reinforcement, three other characteristics of mothers’ modeling were associated with the increase in frequency of imitative responding: amount and variation of the modeling, modeling accompanied by additional vocal communications, and a moderate range in frequency of modeling.

Infant Imitation

287

A second study of imitation explicitly involving reinforcement of infant behavior was conducted by Hursh and Sherman (1973). Using a single-subject multiple-baseline design across vocal responses of 15-24-month-old subjects, Hursh and Sherman investigated the effects of parent-presented models, praise, and repetition of the infant’s vocalizations on the frequency of target infant vocalizations. The combined effects of the three procedures were greater than the individual effects of each one. Also, the effect of the treatment strategies was specific to the target vocalization under treatment. An important point, however, is that the investigators did not consider the infants’ responses to the models to be imitation because they did not meet the definition of generalized imitation described in the introduction to this article. As described in the introduction, generalized imitation refers to generalized responding to untrained as well as trained models. Responding was trained to all the models in the preceding investigation. In a third study of infant imitation involving explicit reinforcement, Poulson and Kymissis (1988) used a multiple-baseline, across-subjects design to evaluate the effects of parent-presented models and praise on motor imitation by 10-month-old normally developing infants. This study demonstrated the effectiveness of parental modeling and praise in evoking infant motor responses that matched a model. Furthermore, the fact that these systematic increases in matching occurred during nonreinforced as well as reinforced trials provides a demonstration of generalized imitation in normally developing infants. A fourth infant imitation study involving reinforcement procedures was conducted by Nunes et al. (1985). Infants at risk for delayed development, who were 9 to 12 months of age, and their low socioeconomic-status teenage mothers, participated as subjects in a single-subject, multiple-baseline design across mother-infant dyads. The independent variable, a parent-training program, consisted of verbal instructions, role-playing episodes, and positive feedback delivered by the experimenter to the mothers. The dependent variables were appropriate parental modeling and matching infant responding. The mothers were trained to produce appropriate models, which were defined as including the following responses: (1) presenting a verbal command, “DO this,” (2) presenting a modeling episode less than 6 seconds in duration, (3) giving the toy to the infant, (4) allowing the infant sufficient time to respond, ( 5 ) prompting the response or repeating the model when the infant failed to match the model, and (6) praising the infant when a matching response occurred. Data on experimenter, parent, and infant responding showed that the introduction of the training program resulted in increased appropriate parental modeling and, correspondingly, in increased matching by the infant. Infant responding in this study may be referred to as generalized imitation because the infants systematically improved in their ability to match

288

Claire L. Pbulson et al.

novel responses and to imitate nonreinforced motor probe responses following the introduction of the parent-training procedures. In summary, a variety of procedures have been used to evoke imitation in infants. Kaye and Marcus used a procedure in which the infant determined the onset of the modeling episode, and Field and her colleagues used a habituation procedure in which offset of infant attending determined the offset of the modeling episode. Such procedures are certainly more likely to ensure the cooperation of the infant with the experimental procedures. All four studies in which infant matching of a modeled response produced consequences presumed to be reinforcing to the infant produced increased infant responding over a baseline comparison condition. In none of the studies was it clearly demonstrated that reinforcement procedures caused an increase in imitation. It would be useful to developmental theorists as well as to practitioners to know the extent to which reinforcement plays a significant role in infant imitation.

VII. Neonatal Imitation of Nonvisible Actions Thirteen studies focused on a very controversial issue: imitation of nonvisible actions (actions the infant does not see himself or herself perform) in the first weeks of life. According to Piaget (1946/1951), imitation of these models should not begin until the fourth stage of the sensorimotor period, between 8 and 12 months of age. This type of imitation cannot occur earlier because of the infant’s inability to coordinate visual schemas with tactilekinesthetic schemas, according to Piaget (1946A951, p. 43). Nevertheless, six studies have been purported to demonstrate that infants can imitate nonvisible movements prior to reaching Piaget’s fourth stage. In the first such study, Maratos (1973) reported that l-month-old infants imitated tongue protrusion and mouth movements. This imitative response decreased over time. In the second study, Meltzoff and Moore (1977) presented four modelsthree facial and one involving finger movements-in random order to infants from 12 to 21 days of age. Each model was presented four times, and all trials were videotaped. For each modeling trial, observers were asked to rank the four actions, from the action they thought the infant was most likely imitating to the action they thought was least likely. The two highest-ranked actions and the two lowest-ranked actions were collapsed for the purpose of analysis. Thus, to the question whether or not the infant was imitating a particular action, the top two ranked responses were counted as “yes” and the bottom two as “no.” The results showed that for each model, the frequency of the matched response (e.g., lip protrusion) was greater than the frequency of each

Infant Imitation

289

nonmatched response (e.g., mouth protrusion, tongue protrusion, and finger movement). The third relevant study, also conducted by Meltzoff and Moore (1977), was a continuation of the investigators’ work described in the previous paragraph. In that work, the experimenter could have repeated each model until the random response of the infant coincided with the modeled behavior. A second experiment was conducted to control for this possible bias. This experiment included two baseline periods: infants using and not using a pacifier. The experimenter, with a passive face, observed the infant and then presented the model (either tongue protrusion or mouth opening) while the infant had the pacifier. Finally, the experimenter resumed a passive face and observed the infant without the pacifier for 150 seconds. The infants emitted the modeled actions more often following the modeling than during the nopacifier baseline, and t he modeled actions occurred selectively relative to periods when a different response was modeled. Anisfeld (1979) and Masters (1979) pointed out two methodological flaws in the first Meltzoff and Moore (1977) experiment. First, a careful examination of the table containing the distribution of infants’ responses across the four models revealed that the infants’ responses that matched the model were more frequent than some responses, but not more frequent than other responses. In addition, if the frequencies of all nonmatching responses were summed, this total would be greater than the frequencies of the putative matching response to the model. Second, when Meltzoff and Moore collapsed the four ranked scores into two, they inadvertently created the possibility that a given response would be judged as imitative equal to 30. That is, observers would be correct during half the trials if they merely guessed which responses had been modeled, independently of any infant responding. An additional problem not raised by Anisfeld and Masters is that infant behavior was not the dependent behavior in this research paradigm. Instead, observers’ guesses as to the nature of the responses of an unseen model was the dependent variable. Considering these methodological limitations, the conclusions one can draw from the Meltzoff and Moore studies are unclear with respect to imitation. The fourth and fifth studies supporting Meltzoff and Moore’s findings were two studies by Field and her colleagues. In the first of these studies (Field etal., 1982), 74 neonates (mean age 36 hours) were exposed to three facial expressions (sad, surprised, and happy face) by an adult model. The models were presented in a counterbalanced order. The observers, who were not aware of which facial expression was being modeled, recorded: (1) total time per trial, (2) predominant target of infant’s visual fixation (eye, mouth, or both), (3) presence of specific mouth movement and eye widening by the infant, (4) presence of relaxed or furrowed brow by the neonate, and ( 5 ) a guess as

290

Claire L. &ulson et al.

to which expression was being modeled by the adult. Comparison of recordings of specific infant eye and mouth movements across the different models showed that infants’ eye widening and mouth opening occurred more frequently following a surprised-face model than other models. Also, a happyface model was associated with infant lip widening, and neonatal protruding lips and furrowed brow occurred more frequently after a sad-faced model. The procedures and data presented in the Field, Woodson, Cohen, Greenburg, Garcia, and Collins (1983) study do not allow as detailed an evaluation, but it appears that similar procedures were used, and similar results were obtained. Some methodological problems occurred in the Field et al. investigations. First, they lacked a baseline control condition, such as that used by Abravanel and Sigafoos (1984); that is, they lacked a measure of the frequency of the targeted neonatal facial expressions emitted spontaneously before the modeling episodes were presented. Second, the frequency of all combined nonmatching infant responses to the happy-face and the sad-face models was greater than the frequency of the matching infant responses to these same models. A sixth study lending support for the possibility of neonatal imitation was conducted by Reissland (1988) with infants during their first hour postpartum in rural Nepal. An experimenter modeled widened lips or pursed lips, alternately, during a modeling condition that was compared with a no-modeling baseline condition for each of 12 infants. Observations of infant lip positions showed that there were significantly more infant lips widened when the model’s lips were widened than when her lips were pursed, and more infant lips pursed when the model’s lips were pursed than when her lips were widened. Further data analysis conducted by the author was reported to reduce the likelihood that the experimenter inadvertently matched the infant’s lip position or that the experimenter merely repeated a given lip position until the infant’s random movements coincided with that of the model. It is unclear whether to describe the modeled responses as static or dynamic, as the experimenter held the modeled mouth positions for an average of 3 to 5 minutes, depending on the attentiveness of the infant. In the Reissland (1988) study, some types of bias might have occurred because the durations of experimental and control conditions could not be held constant, as a result of labor-room procedures. For example, infants might have experienced very short nonmodeling periods and much longer modeling periods, rendering equivocal the reported finding that response frequencies were lower during nonmodeling than modeling conditions. Perhaps the use of a rate-per-minute measure would have been more appropriate than frequency for this and similar comparisons. Nevertheless, if further analysis shows this concern to be unjustified, the Reissland study is one of the least flawed and, consequently, one of the most powerful studies to support

Infant Imitation

291

the phenomenon of neonatal imitation. Reissland noted that five infants received a nonmodeling control condition both before and following a modeling condition. It would be important to know whether the group results were reflected clearly in the data of these five infants. That is, the case for experimental control over neonatal imitation would be strong, indeed, if each of those five infants showed a pattern of low rates of the targeted responding during nonmodeling, followed by high rates of targeted responding during modeling, and followed by low rates again during nonmodeling. In contrast to the foregoing six studies, seven studies failed to show imitation in early infancy. Koepke, Hamm, Legerstee, and Russel (1983) replicated both of the Meltzoff and Moore (1977) experiments. Although they observed that infants’ responding seemed to vary somewhat with the models, they found no evidence of differential responding to the models. This conclusion was also reached by Hayes and Watson (1981) after replicating Meltzoff and Moore’s second experiment. In addition, Hayes and Watson demonstrated that the mouth movement infants were making when they had a pacifier could be used to predict whether they were more likely to open their mouths or protrude their tongues when the pacifier was removed. This finding suggested a possible artifact: In the Meltzoff and Moore (1977) study, the experimenter might have protruded his or her tongue and waited for a push on the pacifier or might have opened his or her mouth and waited for a passive release of the pacifier. In so doing, the experimenter might have biased the experiment to demonstrate the hypothesis of neonatal imitation of nonvisible actions. McKenzie and Over (1983) raised a different issue concerning Meltzoff and Moore’s experiments. McKenzie and Over tested the contingency and structural similarity relationships in adult-infant behavior. One of four gestures (mouth opening, tongue protrusion, arm waving, and hand-to-mouth movement) was performed in front of 1-month-old infants by an experimenter who assumed a neutral expression during the interval between trials. Using a signaldetection analysis, the investigators showed that observers could not determine the modeled action from the infants’ responses beyond chance level. These data suggest that infant responding was at best only grossly tuned to the action of the model. In addition, both frequency and duration measures of infants’ responding indicated that their model matching did not differ significantly from chance expectations. Jacobson (1979) provided data to support a rival hypothesis to that suggested by Meltzoff and Moore’s (1977) first experiment. She demonstrated that at 6 weeks of age, tongue protrusion was as easily evoked by a moving pen and ball as by an adult model. Also, a dangling ring was as effective as an experimenter’s hand in eliciting hand movement in 14-week-old infants. Consistent with this finding are the results of a study by Vinter (1986), who

292

Claire L. fbulson et al.

elicited imitative responding (tongue protrusion and hand opening and closing) in 4-day-old infants presented with dynamic models, but not in infants presented with static models or no models. Heimann and Schaller (1985) found no consistent imitative responding of facial models presented by mothers to their 14- to 21-day-old infants. Nevertheless, they reported that in 6 of the 11 mother-infant dyads, mothers described their infants as having imitated them. The investigators strongly reommended that individual differences among infants be further investigated, as some infants might be high-rate imitators, and others might be low-rate imitators. Kaitz, Meschulach-Sarfaty, Auerbach, and Eidelman (1988) failed to replicate the Field et al. studies with similar infants. Kaitz et al. showed infants the same happy, sad, and surprised models used by Field et al., but they added a fourth model, tongue protrusion, which had been used by Meltzoff in a series of studies of neonatal imitation. Kaitz et al. used Field’s trials-tocriterion procedure for evoking imitation, but only tongue protrusion was reliably produced by the infants in response to the model. The emotional expressions were unaffected by the experimental procedures. Methodological differences between the Field studies and the Kaitz et al. study could account for the differences in findings, as suggested by the latter authors. They suggested that Field’s procedure of recording the pattern of infant eye fixations, in combination with the use of a forced-choice procedure for guessing the nature of the model during trials, might have inadvertently taught the observer to use the infant’s eye fixations to guess the model’s responses. In summary, Meltzoff and Moore‘s (1977, 1983) thesis that neonates exhibit complex cognitive processes that enable them to imitate visible and nonvisible actions is seriously jeopardized by the methodological flaws of their studies, by the difficulty in ruling out rival hypotheses, and by failures to replicate consistently their findings, although Field er al. (1982, 1983) and Reissland (1988) provided positive replication. As Reissland pointed out, it may be significant that the studies showing negative results have all been undertaken with older infants (between 9 and 30 days of age in McKenzie & Over, 1983; mean of 21 days in Koepke et al., 1983; and mean of 48 days in Hayes & Watson, 1981)’ whereas studies with positive findings have all been undertaken with infants no older than a mean of 14 days, such as those in Meltzoff and Moore (1977). In other words, early neonatal imitation may be a different phenomenon from later neonatal imitation. Furthermore, in neither of the preceding uses of the term imitation is the concept of generalized imitation invoked, because the neonatal studies have not involved generalization testing or training procedures. It is clear that we are going t o have to become ever more precise in our definitions of imitation and in our design of experiments that embody those different

Infant Imitation

293

definitions. Indeed, we may be studying different phenomena operating under different principles, rather than any unified mechanism of imitation.

VIII. Past and Future Trends in the Study of Infant Imitation The purpose of most of the studies reviewed here was to describe the development of imitation. Specifically, the investigators proposed to study changes in infant imitative behavior over time. Various theoretical and methodological issues have been raised about these studies. First, most of the studies involved cross-sectional group designs, in which differences between group ages could be attributed to cohort effects, as described, for example, by Porges (1979). Second, when longitudinal designs have been used, the intervals between experimental sessions were at least 1 month. Even in the cross-sectional studies, the age differences among the groups averaged 3 months. If the differences in imitative performance across age were found significant with analysis of variance techniques, the researchers rank ordered the models by difficulty and then used hypothetical constructs such as “cognitive structures” to explain the differences in imitative performance across ages. Considering that during infancy, the rate of change is greater than it is during any other period of life, one could argue that measuring infant performance with 1-month or greater intervals between sessions may not be sufficient to track changes systematically. This procedure may have prevented a more detailed description of important changes occurring across time. Third, in most of the studies, infants demonstrated their responses to a variety of models only a few times. Their responses were then scored as exact imitations, approximate imitations, and not imitations. The studies by Kaye and Marcus (1978, 1981) suggest that such testing and scoring procedures are inadequate because they do not permit observation of successive approximations of features of models across a sequence of trials. Fourth, some of the investigators disregarded the impact of environmental variables on infant’s imitative performance. Notably, however, within the Piagetian perspective, an infant’s continuous interaction with the environment is considered necessary for the construction of the cognitive structures. Thus, both a Piagetian and a behavioral perspective indicate a need to identify the relevant environmental variables and to measure their effects on an infant’s behavior. Some of the reviewed investigators have, in fact, identified a number of such variables. They suggest that an infant’s imitative performance may be affected by factors such as use of objects in gestural imitation (Abravanel et al., 1976; McCall et a/., 1977; Rodgon & Kurdek, 1977), use of socially

294

Claire L. Fbuison et al.

appropriate objects in gestural imitation (Bates ef af.,1980; Killen & Uzgiris, 1981), conditions of an infant’s rearing (Paraskevopoulos & Hunt, 1971), opportunity to interact vocally with parents (Wachs et af., 1971), successive trials with the same model (Kaye & Marcus, 1978; 1981), appropriate pace of model presentation (Waxler & Yarrow, 1975), reinforcement of infant’s responding (Hursh & Sherman, 1973), amount of variation of modeling (Waxler & Yarrow, 1975), additional vocal communication (Waxler & Yarrow, 1975), and infant control of the timing of the model’s presentations (Kaye & Marcus, 1978; 1981). Future research on imitation should help disentangle the variables responsible for the apparent early neonatal appearance, later disappearance, and later reemergence of imitation in infancy. It is clearly possible that the phenomena of imitation are controlled by different processes at different stages in the development of infants. Perhaps early imitation is reflexive in nature, disappearing for awhile, and reemerging later in a more reliable form, as some authors have suggested (e.g., Field er af., 1982; Reissland, 1988). If different processes are, indeed, involved in imitative behavior at different stages in infant development, perhaps the term imitation should be restricted only to some of them. More care in our definition of imitation, and more care in our design of studies invoking the concept of imitation is surely warranted. Some confusion about imitation could be clarified if researchers reserved the term exclusively as a judgment that certain conditions obtained in a study: namely, that the subject’s behavior approximated the model’s behavior because the subject observed the model. Numerous experimental procedures have been used to support the conclusion that imitation has occurred. Some of these were reviewed in the preceding discussion of overall developmental trends and included (1) observation of a control group, (2) inclusion of a baseline control procedure, (3) alteration of the modeling procedure, and (4) measurement of differential frequencies or proportions of individual responses. No studies to date have included all of these procedures, and more sophisticated procedures undoubtedly will be developed. One suggestion for clarifying the processes underlying imitation is that we might profit from experimentally analyzing potential differences among infants of the same age groups, in addition to those of differing age groups. For example, depending on our arrangements of experimental control procedures, it is possible to determine that some infants in a study are imitating and that others are exhibiting matched-dependent behavior, which only appears to be imitation. The sources of experimental control for these two subgroups of infants might then be very different in ways that are directly relevant to our understanding of the processes underlying imitation. It would be useful in identifying such separate cases of experimental control if researchers more often presented the individual-subject data, as well as the grouped

Infant Imitation

295

data, in scientific journals and, indeed, anticipated the need to do so by occasionally using single-subject experimental designs (PouIson & Nunes, 1988). There would then be some studies in which we could draw conclusions about imitation from each individual infant’s data, rather than from the grouped data of a study. To date, no such studies are available. Nevertheless, such studies might facilitate our identification of the processes underlying imitation in individual infants, so that we can examine the generality of those processes through group studies with more confidence that we know what those processes are. ACKNOWLEDGMENTS Financial support for the preparation of this article was made available by Grant HD 22070 from the National Institute of Child Health and Human Development and by Research Award #667441 from the Professional-Staff CongressKity University of New York, Queens College/CUNY.

REFERENCES Abravanel, E., & Gingold, H. (1985). Learning via observation during the second year of life. Developmental Psychology, 21, 614-623. Abravanel, E., Levan-Goldschmidt, E., & Stevenson, M. B. (1976). Action imitation: The early phase of infancy. Child Development, 41, 1032-1044. Abravanel, E., & Sigafoos, A. D. (1984). Exploring the presence of imitation during early infancy. Child Development, 55, 381-392. Anisfeld, M. (1979). Interpreting “imitative” responses in early infancy. Science, 205, 214-215. Baer, D. M., & Deguchi, H. (1985). Generalized imitation from a radical-behavioral viewpoint. In S . Reiss & R. Bootzin (Eds.), Theoretical issues in behavior therapy (pp. 179-217). New York: Academic Press. Baer, D. M., Peterson, R. F., &Sherman, J. A. (1967). The development of imitation by reinforcing behavioral similarity to a model. Journal of Experimental Analysis of Behavior, 10, 405-416. Baer, D. M., & Sherman, J. A. (1964). Reinforcement control of generalized imitation in young children. Journal of Experimental Child Psychology, 1, 31-49. Bandura, A. (1965). Influence of models’ reinforcement contingencies on the acquisiton of imitative responses. Journal of Personality and Social Psychology, 1, 589-595. Bandura, A., & Huston, A. C. (1961). Identification as a process of incidental learning. Journal of Abnormal and Social Psychology, 63, 311-318. Bates, E., Bretherton, I., Shore, C., Snyder, L., & Volterra, V. (1980). Vocal and gestural symbols at 13 months. Merrill-Palmer Quarterly, 26, 407-423. Coates, B., & Hartup, W. W. (1969). Age and verbalization in observational learning. Developmental Psychology, 1, 556-562. Egel, A. L., Richman, G. S., & Koegel, R. L. (1981). Normal peer models and autistic children’s learning. Journal of Applied Behavior Analysis, 14, 3-12. Fenson, L., & Ramsay, D. S. (1981). Effects of modeling action sequences on the play of twelve-,

296

Claire L. Poulson et al.

fifteen-, and nineteen-month-old children. Child Development, 52, 1028-1036. Field, T., Goldstein, S., Vega-Lahr, N., & Porter, K. (1986). Changes in imitative behavior during early infancy. Infant Behavior and Development, 9, 415-421. Field, T. M., Woodson, R., Cohen, D., Greenberg, R., Garcia, R., & Collins, K. (1983). Discrimination and imitation of facial expressions by term and preterm neonates. Infant Behavior and Development, 6, 485-489. Field, T. M., Woodson, R., Greenberg. R., & Cohen, D. (1982). Discrimination and imitation of facial expressions by neonates. Science, 218, 179-181. Fontaine, R. (1984). Imitative skills between birth and six months. Infant Behavior and Development, 7, 323-333. Garcia, E., Baer, D. M., & Firestone, I. (1971). The development of generalized imitation within topographically determined boundaries. Journal of Applied Behavior Analysis, 4, 101-1 12. Gratch, G.. & Landers, W. F. (1971). Stage IV of Piaget’s theory of infants’ object concepts: A longitudinal study. Child Development, 42, 359-372. Guillaume, D. (1971). Imitation in children. Chicago: University of Chicago Press. (Original work published in 1926) Hallahan, D. P., Kauffman, J. M., Kneedler, R. D., Snell, M. E., & Richards, H. C. (1977). Being imitated by an adult and the subsequent imitative behavior of retarded children. American Journal of Mental Deficiency, 81, 556-560. Hartup, W. W., & Coates, B. (1970). The role of imitation in childhood socialization. In R. A. Hoppe, G. A. Milton, & E. C. Simmel (Eds.), Early experiences and theprocesses of socialization (pp. 109-142). New York: Academic Press. Hayes, L. A., & Watson, J. S. (1981). Neonatal imitations: Fact or artifact? Developmental Psychology, 17, 655-660. Heimann, M., & Schaller, J. (1985). Imitative reactions among 14-21 day old infants. Infant Mental Health Journal, 6 , 31-39. Hetherington, E. M. (1975). A developmental study of the effects of sex of the dominant parent on sex role preference, identification, and imitation in children. Journal of Personality and Social Psychology, 6, 119-125. Hewett, F. M. (1965). Teaching speech to an autistic child through operant conditioning. American Journal of Orthopsychiatry, 35, 927-936. Hursh, D. E., & Sherman, J. A. (1973). The effects of parent-presented models and praise on the vocal behavior of their children. Journal of Experimental Child Psychology, 15,328-339. Jacobson, S. W. (1979). Matching behavior in the young infant. Child Development, 50,425-430. Kaitz, M., Meschulach-Sarfaty, O., Auerbach, J., & Eidelman, A. (1988). A reexamination of newborns’ ability to imitate facial expressions. Developmental Psychology, 24, 3-7. a y e , K., & Markus, J. (1978). Imitation over a series of trials without feedback: Age six months. Idant Behavior and Development, 1, 141-155. Kaye, K., & Marcus, J. (1981). Infant imitation: The sensory-motor agenda. Developmental Psychology, 17, 258-265. Kessen, W., Levine, J., & Wendrich, K. A. (1979). The imitation of pitch in infants. Infanf Behavior and Development, 2, 93-99. Killen, M., & Uzgiris, I. C. (1981). Imitation of actions with objects: The role of social meaning. Journal of Genetic Psychology, 138, 219-229. Koepke, J. E., Hamm, M., Legerstee, M., & Russel, M. (1983). Neonatal imitation: TWO failures to replicate. Infant Behavior and Development, 6 , 97-102. Konorski, J., & Miller, S. (1937). On two types of conditioned reflex. Journal of Genera/ Psychology, 16, 264-272. Kornhaber, R, C., & Schroeder, H. E. (1975). Importance of model similarity on extinction of avoidance behavior in children. Journal of Consulting and Clinical Psychology, 43,601-607.

Infant Imitation

297

Largo, R. H., & Howard, J. A. (1979). Developmental progression in play behavior of children between nine and thirteeen months. I: Spontaneous play and imitation. Developmental Medicine and Child Neurology, 21, 299-310. Lovaas, 0.1. (1977). The autistic child: Language development through behavior modification. New York: Irvington. Lovaas, 0. I., Berberich, J. P., Perloff, B. F., & Schaeffer, B. (1966). Acquisition of imitative speech by schizophrenic children. Science, 151, 705-707. Lovaas, 0. I., Freitas, L., Nelson, K., & Whalen, C. (1967). The establishment of imitation and its use for the development of complex behavior in schizophrenic children. Behavior Research and Therapy, 5, 171-182. Maratos, 0. (1973, April). The origin and development of imitation in thefirst six months of life. Paper presented at the meeting of the British Psychological Assocation, Liverpool. Martin, J. A. (1971). The effect of incongruent instructions and consequences on imitation in retarded children. Journal of Applied Behavior Analysis, 5, 467-475. Masters, J. (1979). Interpreting “imitative” responses in early infancy. Science, 205, 215. McCabe, M. A., & Uzgiris, I. C. (1983). Effects of model and action on imitation in infancy. MerriN-Palmer Quarterly, 29, 69-82. McCall, R. B., Parke, R. D., & Kavanaugh, R. D. (1977). Imitation of live and televised models by children one to three years of age. Monographs of the Society for Research in Child Development, 42 ( 5 , Serial No. 173). McKenzie, B., &Over, R. (1983). Young infants fail to imitate facial and manual gestures. Infant Behavior and Development, 6, 85-95. Meltzoff, A. N. (1985). Immediate and deferred imitation in fourteen- and twenty-four-monthold infants. Child Development, 56, 62-72. Meltzoff, A. N. (1988a). Imitation of televised models by infants. Child Development, 59, 1221-1229. Meltzoff, A. N. (1988b). Infant imitation after a I-week delay: Long-term memory for novel acts and multiple stimuli. Developmental Psychology, 24, 470-476. Meltzoff, A. N. (1988~).Infant imitation and memory: Nine-month-olds in immediate and deferred tests. Child Development, 59, 217-225. Meltzoff, A. N., & Moore, M. K. (1977). Imitation of facial and manual gestures by human neonates. Science, 198, 75-78. Meltzoff, A. N., & Moore, M. K. (1983). Newborn infants imitate facial gestures. Child Development, 54, 702-709. Miller, N. E., & Dollard, J. (1941). Social learning and imitation. New Haven, CT: Yale University Press. Moran, G., Krupka, A., mtton, A., & Symons, D. (1987). Patterns of maternal and infant imitation during play. Infant Behavior and Development, 10, 477-491. Nelson, G. L., Cone, J. D., & Hanson, C. R. (1973). Training correct utensil use in retarded children: Modeling vs. physical guidance. American Journal of Mental Deficiency, 80, 114-122. Newson, J. (1974). Toward a theory of infant understanding. Bulletin of the British Psychological Society, 27, 251-257. Nunes, L. R. P., Poulson, C. L., Nunes, F., Almeida, M., & Warren, S. F. (1985, May). Increasing accuracy of idant imitation as a function of appropriate modeling by disadvantaged teenage mothers. Paper presented at the meeting of the Association for Behavior Analysis, Columbus, OH. Paraskevopoulos, J., & Hunt, J. McV. (1971). Object construction and imitation under differing conditions of rearing. Journal of Genetic Psychology, 119, 301-321. Parton, D. A. (1976). Learning to imitate in infancy. Child Development, 47, 14-31. Parton, D. A., & Priefert, J. J. (1975). The value of being imitated. Journal of Experimental

298

Claire L. Poulson et ai.

Child Psychology, 20, 286-295. Pawlby, S. F. (1977). Imitative interaction. In H. R. Schaeffer (Ed.), Studies in mother=infant interaction (pp. 203-224). London: Academic Press. Piaget, J. (1951). Play, dreams, and imitation in childhood. New York: Norton. (Original work published in 1946) Porges, S. W. (1979). Developmental designs for infancy research. In J. D. Osofsky (Ed.), Handbook of infant development (pp. 742-765). New York: Wiley. Poulson, C. L., & Kymissis, E. (1988). Analysis of generalized motor imitation in infants. Journal of Experimental Child Psychology, 46,324-336. Poulson, C. L., & Nunes, L. R. P. (1988). The infant vocal conditioning literature: A theoretical and methodological critique. Journal of Experimental Child Psychology, 46, 438-450. Reissland, N. (1988). Neonatal imitation in the first hour of life: Observations in rural Nepal. Developmental Psychology, 24, 464-469. Rodgon, M. M., & Kurdek, L. A. (1977). Vocal and gestural imitation in 8-, 14-, and 20-monthold children. Journal of Genetic Psychology, 131, 115-123. Steinman, W. M. (1970). The social control of generalized imitation. Journal of Applied Behavior Analysis, 3, 159-167. Striefel, J. A., & Phelan, J. G. (1972). Use of reinforcement of behavioral similarity to establish imitative behavior in young mentally retarded children. American Journal of Mental Deficiency, 17, 239-241. Terrace, H. W. (1963). Discrimination learning with and without “errors.”Journal of Experimental Analysis of Behavior, 6 , 1-27. Tulkin, S. R., & Kagan, J. (1972). Mother-child interaction in the first year of life. Child Development, 43, 31-41. Uzgiris, I. C. (1973). Patterns of vocal and gestural imitation in infants. In L. J. Stone, H. T. Smith, & L. B. Murphy (Eds.), The competent idant @p. 599-604). New York: Basic Books. Valentine, C. W.(1930). The psychology of imitation with special reference to early childhood. British Journal of Psychology, 21, 105-132. Vinter, A. (1986). The role of movement in elicitingearly imitations. Child Development, 57,66-71. Wachs, T. D., Uzgiris, I. C., & Hunt, J. McV. (1971). Cognitive development in infants of different age levels and from different environmental backgrounds: An exploratory investigation. MerriII-Palmer Quarterb 17, 283-317. Watson, M. W., & Fischer, K. W. (1977). A developmental sequence of agent use in late infancy. Child Development, 48, 828-836. Waxler, C. Z., & Yarrow, M. R. (1975). An observational study of maternal models. Developmenral Psychology, 11, 485-494. Yando, R., Seitz, V., & Zigler, E. (1978). Imitation: A developmental perspective. Hillside, NJ: Erlbaum.

AUTHOR INDEX Numbers in italics refer to the pages on which the complete references are cited.

A

Bassert. E. 222. 226, 233. 234. 239, 243 Bates. E.. 294, 2Y5 Bearison. D. J..97. 112. 113, 114. 115. 118. 137 220, 241 Bee, H. L.. 91, 137 Beehe. S.. 220. 245 Behrend. D. A , . 107. 1 1 1 . 137. 142 Behrend. S . , 97. 112. 113. 114. 141 Beilin. H . , 174. 195, IY8 Belknap. B . , 218. 245 Berherich. J. P., 773. 207 Uerger. J . , 194. 1%’ Berndt. T. J . , 119, 137 Bernstein. A,. 3Y Bersani, C.. 209. 243 Bertenthal. B. I.. 250, 268 Bertin. J . . 148. 162, I Y H Bialystok. E.. 223. 233, 241 Bijou. S . W.. 295 Billman. D. 0.. 17. 20. 38 Blades. M . , 165. 167. 170, 172, IYK Blaut, A. S . . 164. 165, 166. 167. 16X. 169. 170. 172. 173, 190. lY8 Blaut. J . M.. 165. 166. 167. 168. 169. 170. IYX. ~

Ahel. R . R . . 164. 1% Aboud. F., 103, 119. 141 Ahravanel. E . , 238. 244, 276. 277. 279. 280, 281. 282.283, 286.290.293, 2YS Ackerman, A. M . , 227. 242 Acredolo, L. P.. 3, 9, 21. 26. 36, 152, 153. 155. 156, 158. 159, 160. 161. 162. 164, 167. 170. 172. 175. 176. 191. IYK. 19Y. 207. 215,241 Adamson. L. B.. 90. 140 Aehli. H.. 208. 220. 241 Ainslie. R . C.. 121. 136 Alek. M.. 253,256,257,258. 269 Allen, G. L.. 9, 10. 23. 25, 39 Allen. V. L . , 113. 115. 136 Almeida. M . . 275, 281. 287, ZYX Ames. G. J . . 114, 116, 137 Anderson, E.. 5 0 , 86 Anisfeld, M . . 289, ZYY A m . E , 93. 107. 144 Any. J . , 253, 256. 257. 258. 269 Ashworth. P. 0..92, 137 Aucrbach. J . . 292. 300 Ayers-Lopcz, S.. 114, 138 Azmitia. M.. 112. 113. 117. 120, 130. I37

-701

Blom. D. E..99. 100, 115. I I X , 144 Blueatein, N.. 3 . 26. 36. 152, 153. 155, 156. 15X. 159. 160. 162. 164. 167, 170, 172, 175, 176. 191. 108 Bogartz, R . S . , 251, 268 Boies. S.. 251. 254. 259. 269 Bolsmier. J. D.. 224. 245 Bopp, M . , 220. 245 Borke. H.. 223, 225. 226. 231. 241 Botkin. P. T.. 109. 113. 13Y. 221, 24.3

B Baer, D. M . . 272. 273, 2Y5 Batllargeon. R . . 210, 243 Ball, T. M , 151, IYY Bandura. A,. 98, 99. 125. 127, 137. 273, 2Y5

299

300

Aurhor Index

Botvin. G . J.. 98. 100. 114. 118. 137 Bowd, A , . 218, 241 Boyes. M . . 44. 86 Brddy. J . E., 103, 119, 141 Braine. L., 208. 229. 241 Brainerd, C. J . . 206, 246 Braniel. 117. 137 Bransford. J. D., 42. 85. 86. 101, 110, 137 Breniner, J. G., 207, 241 Brenner. 1.. 106. 137, 141 Bretherton. I.. 294. 295 Broadbent. D. E., 254, 268 Brodzinsky, D. M.. 210. 218,221. 228, 24/ Brooks, V.. 25. 38 Brassard. A , , 92. 96. 114, 14/ Broughton. J.. 86 Brown. A . L.. 3, 10, 17, 18, 19, 36. 37. 39.42, 8 5 . 86, 94. 101, 110. 132. 137. 177, 198 Brown. E., 104. 106. 137 Brownell. C. A . , 104. 106, 111, 130, 137 Bruner, A . L., 85.86 Bruner, J. S., 92, 94, 95. 108. 110. 137, 143, 144, 161. 198 Bryant, N . R., 137. 94 Bryant, P. E.. 215, 241. 247 Budwig. N . A , . 95, 108, 144 Bukatko. D., 28, 336

c Campione. J . C.. 42.85, 86,94, 101. 110. 137 Cantor, G. N., 100. 137. 250, 2.51. 256, 267, 268, 270 Cantor, J. H., 268 Carey, R. N . . 228.229. 245 Carey, S., 244 Caspy, T., 257, 258, 269 Cassel, T., 137 Cassidy, D. J . , 3, 10, 37 Cavanaugh, J. C., 111, 116. 138 Cazden. C. B., 94, 104. 113, 114. 118, 120, 135, 138. 139

Cecil, L. S . . 257, 268 Chandler. M. J., 44, 83, 86 Charlesworth. W. R . , 119, 140 Chi, M. T. H., 3, 36 Cicerelli, V. G . , 121, 122, 1.18 Coares. B., 273,274.296 Cohen. D., 281, 286.289, 290. 292, 294, 296

Cohen, L. B., 250, 268 Cohen, P. A , . 117, 138 Cohen. R., 152. I98 Cohen, S. R., 3. 36 Coie, J. D., 226, 229. 242 Cole. M . , 28. 36, 94. 113, 138. 142 Collins, K . . 286, 290, 296 Condon, W. S , 9 3 , I38 Cone, 1. D., 273. 301 Connolly, J., 120, 138 Cooper. C. R.. 112. 114, I38 Cooper, L., 85, 87 Cooper, P., 90, 141 Corbitt, R . , 8 5 , 87 Costanzo, P. R., 226,229.242 Cowan, P. A , , 218, 243 Cox, M . V.. 219. 223,224,229,230. 233. 238, 242 Cox, M . , 113, 138 Crisafi, M. A , , 17. 36 Croft, K . , 209, 210, 243

D Daehler. M. W., 28, 36 Daggs. D. G., 150. 199 Dale, P. S . , 94, 138 Damon. W., 115. 119, 131, 141, 138 Ddrvizeh. Z . , 167, 168, 169, 175, 176, 200 Davis, A., 110, 138 Day, R. H . , 21, 38,207, 245 Dayton. C. M . , 227, 242 De Lisi. R . , 221, 237, 238, 242 Dean. A. L., 240,242 Deguchi, H., 272, 273, 295 DeLoache, J. S., 3, 10, 14, 18.25.28. 36,37, 107, 108. 138, 174, I98 Dennett, D. C.. 5 8 , 86 Dickinson, D., 114, 138 Dirks, J., 25, 37 Dodgson, C.. 160, 198 Doise, W., 90.92, 96, 114, 112, 115, 116, 118, 138, 141,220. 242, 245 Dollard, J . . 98. 125, 138, 141,272,280, 297 Donaldson, M . , 109, 113, 138. 209, 219, 242, 244 Downs, R . M., 26. 27, 37, 148, 149. 150, 162, 173. 175, 178, 181, 182, 186, 198. IY9, 200

Author Index

Doyle, A . , 120, 138 Dubas. J. S . , 217, 242

E Eastman, J. R., 165, 199 Echols, C. H., 18, 36 Eckert, H. M., 35,37 Eder, R. A., 107, 138 Eder. R . , 208, 229, 241 Edwards. K., 215, 246 Egel, A. L., 273, 296 Eidelman, A., 292, 296 Eiser. C., 223, 226, 227, 242 Ekman, G., 165, 199 Elder, J. L., 31. 37, 38 Eliot, J.. 227, 242 Ellis, S . , 90, 92, 94, 96. 105, 110, 112, 118, 138. 139, 142 Enright, R. D.. 222, 221, 242, 244 Ensing, S . S . , 223,224,225,226.228.231, 246 Epstein, S., 205, 242 Ericsson, K. A , , 58, 86 Espenschade. A . S . , 35, 37 Estes, D.. 46.47,48,49,50,56,80,87,207, 241 Everett, B. A . , 209, 210, 243

F Farkas, M. S . . 257,261, 263,264, 269 Farnill. D., 226, 229. 242 Fehr, K . M., 223, 242 Fehr, L. A., 204,219, 222,223, 227, 242, 244 k i n , G. G.. 152, 161. 200 Feldman, A , , 164, 199 Feldman, D. H., 151. 199 Fenson, L , 277, 278, 296 Ferrara, R . A , , 42, 85, 86, 94, 101, 110. 132, 137 Feurstein, R., 94, 114, 132, 139 Field, D., 112, 139 Field, T. M., 276,281,285,286. 289, 290, 292,294,296 Filardo, E. K., 97, 112, 113, 115, 118. 137, 220. 241 Finlay, D. C . , 239, 243

301

Finley, G. E., 218, 243 Firestone, I . , 273, 296 Fischer, K. W., 101, 139, 278, 280, 282, 298 Fishbein, H. D., 222,223,228,229,231,232, 242,243,245 Flavell, E. R., 33, 37, 83, 86, 208, 209, 243 Flavell, J. H., 33, 35, 37.42, 83, 86, 109, 113, 139, 208, 209, 210, 218, 221, 232, 243, 244, 245, 246,250,268 Fleckie, M. S . , 91, 137 Flesson, J., 103, 143 Fontaine, R . , 276, 296 Ford, M. E., 206, 243 Forman. E. A . , 94, 114, 118, 120, 135, 139 Frank, E. K., 152, 161, 200 Freeman, N. H., 3, 37 Freitas, L., 273, 297 Fry. C. L., 109, 113, 139, 221, 243 Furman, W., 106, 120, 139

G Gallistell, R . , 21, 37 Garcia. E . , 273, 296 Garcia, R., 286,290,292. 296 Gardner, H . , 153, 159, 184, 199,201 Gardner, W., 104, 107, 108, 109, 110, 142 Garner, J . , 220, 243 Garton, A. F., 114, 115, 118, 142 Garvey. C., 130, 139 Gauvain, M . , 90,92, 108, 112, 113, 139, 142, 152, 199 Geiringer, E., 98, 143 Gelman, R., 113, 139, 210. 243 Genova, P., 110, 143 Gentner, D., 3, 15, 18, 19, 20, 27, 37 Gerniond, J., 106, 142 Gewirtz, J. L . , 98, 139 Getz, M., 238, 243 Gibson. E., 25, 37 Gick, M. L., 16, 17, 18, 37, 110, 139 Gilbride. K . , 109, 142 Gingold, H., 279, 295 Glachan, M . , 115, 116, 139 Gleitman, H . , 152, 158, 159, 170, 199, 216, 244 Click, J., 133, I39 Golbeck, S. L., 238, 245 Goldberg, M . H., 119, 139 Goldstein. S., 276, 281, 286, 296

302

Author Index

Gollin, E. S., 221,224, 246, 247 Goncu, A , , 109, I39 Goodsitt, J.. 95, 107, 139 Gottman, J. M., 119, 139 Grady-Reitan, J., 95, 107. 139 Gratch, G . , 282, 296 Graves, Z. R., 133, 139 Green. F. L., 33, 37, 83, 86, 208, 243 Greenberg, R., 281, 286.289,290,292,294, 296 Greenburg, M. T., 42, 87 Greenfield, P. M., 85, 86 Greenfield, P., 108, 139 Gruendel, J. M., 3 , 38 Guberman, S . R., 90, 108, 142 Guillaume, D., 272, 274, 275, 296 Gullo, D. E., 209, 243 Gumerman. R. A , , 123, 141 Gzesh, S. M., 218,223,228,229, 243

H Hall, E., 267, 268 Hallahan. D. P.,273, 296 Hamm, M., 291,297 Hanson, C. R . , 273. 297 Hardwick, D., 234, 243 Harris. L. J., 243 Harris. P. L., 222,226,233, 234, 239, 243 Harrison, N., 169, 175, 176, 200 Hart, R., 152, 199 Hartley, D. G . , 28, 37 Hartup, W. W., 103, 111, 119, 139, 140, 141, 273,274, 296 Harvey, P. D. A , , 147, 199 Harvey, W. O., 240,242 Hawkins, J., 114, 115, 127, 131, 140 Hayes, L. A,, 296 Hazen, N. L., 152, 199 Heide. P.. 114, 115, 131, 140 Heimann, M., 277, 292, 296 Helm, D., 83, 86 Herman, J. F., 215, 238, 243 Hess, R. D., 91, 140 Hetherington, E. M . , 296 Hewett, E M., 273, 296 Higgins, E. T., 25, 38 Hirtle, S . C . , 151, 199 Hjertholm, E., 95, 140

Hobson, R. P., 209, 210, 219, 244 Hochberg, J., 25, 38 Hockaday, F., 119, 140 Hogrefe, G . J., 33, 38, 83,87 Holyoak, K . J.. 15, 16, 17, 18, 20, 37, 38, 110, 139 Homolski, M., 114, 115, 131. 140 Hooks, J., 224,245 Horan, P. E , 223, 224,225,226,228.231, 244,246 Horn, C. C., 262.268 Howard, J. A,, 278,297 Hoy, E. A . , 229, 233, 244 Hughes, M., 209, 219, 244 Hunt, I. McV., 282, 283, 294, 298 Hursh, D. E., 275,287, 294. 296 Hurwitz, S., 151, 200 Huston, A. C., 277, 295 Huttenlocher, J . , 25, 38, 210, 215, 217, 218, 221,222,224,225,231,232, 233,239, 244 Huxley. A . , 146, 199

I Ihrig, L. H., 215, 246 Ihsen, E., 207, 245 Inhelder, B., 10,38, 85, 87, 151, 174, 186, 200, 204,210, 211, 212, 213, 214, 215, 217, 230, 236,238,239, 245 Ives, W., 234, 244

J Jackson, J. P., 221, 228, 241 Jacobsen, T. L., 218, 223, 228, 229,230, 244 Jacobson, S . W., 276. 291, 296 Jaffe, J., 93, 140 Jamison, W., 238, 247 Jankovic, I. N., 151, 199 larvis, P. E., 109, 113, 139, 221, 243 Jeffery, W. E . , 250, 268 Jensen, M. R., 94, 114, 132, 139 Johnson, C. N., 83, 86 Johnson, D. W., 114. 115, 140, 143 Johnson, J., 121, 140 Johnson, N. S . , 3, 38 Johnson, R. T., 114, 115, 140, 143

Author lndex

Johnston. J. M.. 296 Jones. C. P 90. 140 Joseph. L. J., 253, 256, 269 Junn. E. N., 17, 20. 38 Jurkovic. G. J . , 85. 87

.

303

Kunst-Wilson, W. R.. 267, 269 Kuo. F., 97, 112. 113. 114. 141 Kurdek, L. A . . 277. 279, 280, 293, 298 Kymissis, E., 275, 281, 287, 298

L K Kabane. D. C., 222. 246 Kagan. 1.. 129. 140.283. 284. 298 Kagan. S., 218. 244 Kaitz. M . , 292, 296 Kane, M . J., 18, 19, 36 Karmiloff-Smith, A., 3, 38 Kauffman, J. M . , 273, 296 Kavanaugh. K. D . , 276. 277, 279, 280, 281. 293, 297 Kaye. K . . 285. 288. 294. 296 Kearsley. R. B.. 25.39 Keating, M . B . , 21. 38 Keiffer. K.. 228, 229. 231. 232, 243 Keil, E C.. 42, 43,44, 46.48, 86 Kellman. P. J.. 269, 250. 269 Ken; N . H., 85.87 Kerwin. S . , 90. 141 Kessen. W., 279, 296 Kielgast. K.. 224. 244 Killen. M . , 140,278, 294, 297 Kirasic. K. C.. 9. 10. 23, 25. 39 Kneedler, R. D., 273, 296 Knudson. K . H . , 218, 244 Koegel. R . L., 273, 296 Koepke. J . E., 291, 297 Koester, L. S . . 121, 140 Kohlberg, L., 95. 140 Kolstad, D. V. 14. 37 Konorski, J . . 273. 297 Kontos. S . , 95. 107, 110. 140 Kornhaber, K. C., 297 Kosslyn. S. M., 2.38. 151, 199 Krappman, L.. 105, 119, 140 Kraut, A . G . , 255,257,258, 259. 260, 261, 263,264.265, 268. 269 Krupka, A . . 284, 285, 297 Ksir, C. J . , 257, 269 Kuhn. D., 90,96,98, 110, 128, 130, 140, 141 Kulhavy, R. W., 164, 198 Kulik, C. D., 117, 138 Kulik, J. A . , 117, 138 I

La Freniere, P. J . , 119, 140 LaBerge. D., 262. 269 Laboratory of Comparative Human Cognition, 90.92. 101, 108, 110. 140 Landau. B., 152, 156, 157, 158. 159, 160. 170, 171, 172. 173. 192. 199, 216. 244 Landers, R. 20. 37 Landers, W. F., 282. 296 Lange, G., 110. 138 Laosa. L. M., 90, 92, 140 Lapsley, D. K., 222, 227, 242. 244 Largo. R . H.. 278,297 Larsen. G . Y., 238. 244 Lasky, R . E.. 240, 244. 259. 269 Latham. C.. 208, 209. 210, 243 Laurendeau, M . , 10, 14, 15, 38,56.78,79,80, 81, 87. 215, 225,230, 236, 237. 244 Laursen, B., 119, 140 Lave, J.. 108, 101, 139, 142 Leekam, S. R.. 33,34, 35, 38 Legerstee, M . , 291, 297 Lempers, J. D., 209. 244 Leslie, A. M . . 38 Leuba, C., 125, 140 Levan-Goldschmidt. E.. 276. 277, 280, 281. 282,283,293,295 Levine. J., 279, 296 Levine, M., 151, 199 Lewis, S . , 228.229.231.232.243 Liben, L. S.. 3 , 26, 27, 37, 38, 150, 151. 152. 158, 170, 173. 178. 181, 182, 186, 199, 200,216, 218, 228, 229.238. 244. 245 Light, P., 115, 116, 139, 218, 245 Lindauer, B., 90, 141 Lindmdn, R . , 165, 199 Lkelgast. K., 244 Lobeck, A. K., 161, 200 Locker, R.,221, 237.238. 242 Lockman, J. J., 152, 156, 199, 200 Lonardo. R., 28. 36 Lovaas. 0. I . . 273, 297 Lovell, K., 215, 245

Aurhor Index

304

Lubow, R. E.. 253,256,257,258. 269 Luria, A . R . , 90.94, 141 Lyons. W.. 84. X7

M Maccoby, E. E., 119. 139 MacWhinney. B., 84. 87 Magzamen. S . , 97, 112, 113, 115, 118, 137. 220. 241 Malkin, C., 109. 142 Mandler. J. M . . 2, 3, 38. 158, 200. 214. 245 Manis, F. R . , 262, 268 Maratos. O., 276,279,288. 297 Marcus. C. B.. 122, 144 Markman. E. M . , 67. 87 Markus, J., 285,288,294, ZY6 Marnior. G . S . , 85, 87,239. 245 Marquis, A.. 114, 1.361 Martin. I. A.. 273, 297 Marvin. R. S . , 42, 87 Masangkay, Z . S . , 208.209, 245 Mascolo, M. F., 151. 199 Masters, J.. 289, 297 Malthews. W. S . . 220. 245 Maynard, J . . 25, 37 Mazzeo, J.. 223. 224, 225. 226, 228, 231, 246 McCdhe, M. A . . 277,278. 281. 297 McCall, R. B. 276, 277. 279.280. 28 1,293. 297 McCleary. G . S. Jr.. 165, 166, 167, 168, 169, 170. 198

McCluskey, K. A , , 208, 209. 24s Mclntyre, C. W.. 208.209. 234, 243. 245 McKenzie. B. E . , 21,3X, 207. 245. 291, 292, 297 McKoon. G . , 151. 200 McLane. J. B., 90. 95. 108, 114, 141, f44 McMahon, P. M., 223, 227.242 McNaniam. T. P., 151, 200 McNamee, G. D.. 90, 95, 108, 141, 144 Mead. G. H., 92. 95, 100, I41 Meltzotf. A. N . , 279,288, 289,291. 292. 297 Meschulach-Sarfaty, O., 292, 296 Metzler, J.. 246, 217 Meyers. W. J.. 253. 256, 269 Middleton. D. J.. 108, I44 Millar. S., 158, 170, 200

Miller, D. 1.. 250, 269 Miller, J. W.. 224, 230, 245 Miller. N. E., 98. 125, 13X. 141, 272, 280, 297 Miller, R . , 94. 114, 132. 139 Miller, S., 273, 297 Minick, N., 93, 107. 144 Minnigerode, F. A . . 228, 229, 245 Miranda, C.. 238, 243 Mischel, W., 206, 242 Misciones. J. L., 42, 87 Mistry, J . , 106, 142 Monmonier. M. S . , 148. 194, 200 Moore. M. K.. 288,289, 291. 292. 297 Moran, G.. 284.285, 297 Morgan. A . . 106, 142 Morrison. F. J., 263, 267, 269 Morrison. H . , 130, 14/ Morrison. J. L . , 263, 200 Morss, J. R.,207, 245 Moss, E. B., 111. 141 Mowrer, 0. H.. 98, I41 Muehrcke, P. C.. 163. 200 Mueller. E.. 106, 137. 141 Mugny. G . . 90,92.96. 112, 114, 115. 116, 118, 138. 141,220,242, 245 Muller, A . . 97, 112, 113, 114. 141 Murphy, C. M . . 26,38 Murray, E 8..98. IOU, 114, 115, 116, 118, 137. 141

Murray. J. P., 96. 112, 115, 116, 141

N Nelson. G . L..273. 297 Nelson, J . , 103, 119, 141 Nelson. K., 3.38.90. 141, 273. 2Y7 Nelson-Lecall, S. A , . 123. 141 Newcomb, A . F.. 103, 119, 14I Newcornbe, N.. 3. 38. 152. 200, 204.210.215, 216,217,218, 222,234, 239. 244. 245

Newson. J., 284, 298 Nigl, A. .I.. 223.228. 229,232. 245 Ninio. A . , 95. 107. 108, 141 Nisbett. R. E.. 58. 84, 87 Nix, C., 218, 245 Nunes, F., 275,281. 287, 2YX Nunes, L . R . P., 275, 281. 287. 294. 298 Nyrnan, B. A., 91, 137

Author Index

0 O'Brien. R G ,42, 87 O'Leary. K . 2 5 , 3 9 Obgolu, 0 A , 218. 245 Ologbaiye. 0 0 218, 245 Olsen, M G , 215. 241 Olver. R R , 85. 86 Onianwn, R C , 208, 209. 210, 243 Oswald, H , 105, 119. 140 Over. R , 391. 292. 297 Overton, W F , 221, 228, 241

.

P Palij. M . , 151, I99 Palmer, S . E.. 2. 38 Paraskevopoulos. J.. 282. 294. 2Y8 Parib.S . . 90. 141 Parke. R. D.. 276,277. 279.280. 281, 293. 297 Parkhurst, J. T., 119%139 Parks. C., 107. 138 Parton. D. A , . 273. 298 Paterson. R F.. 272. 273. 297 Patterson, A . H.. 3, 38. 152. 200 Pawlby, S . F.. 284, 298 Pederson. D. R . , 31. 37. 38 Pennypacker. H. S . , 297 Perlmutter. M.. 95, 97, 101, 107, 111, 112. 113. 114. 116, 137, 138. 139. 111. 142 Perlot'f. 8. F.. 273. 297 Perner. J.. 33. 34, 35. 38, 39 Perret-Clermont. A . N . , 92, 96. 97. 98, 99. 112, 113. 114. 115. /3X. 141, 220. 242 Perry. J. C.. 93. 140 Perry. M. D.. 153, 162, 163. 164, 175, 184. 193, 194. 200 Phelan. J. G . , 273, 29X Phelps. E . , 90. 110, 128. 131. 140, 141 Piaget. J.. 2. 10, 3X. 42. 43. 44. 48. 56. 58. 67, 68.77,78,81.82,85. 87.96.99. 100. 101. 104. 127. 142. 151. 174. 178. 186. 200. 204. 206. 210. 211. 212. 213. 214. 215. 217. 230. 336. 238. 239, 240, 245. 272. 274. 275. 282.288. 298 Pick. H. L.. Jr., 152, 156, 199. 1100, 215, 234.

241.243 Pillow, B . H.. 209, 210. 218. 246

305

Pinard, A., 10, 14, 15. 38. 56. 78. 79. X0, 81. X7.215, 225. 230. 236,237, 244 Plaetzer. B., 37, 107. 108. I38 Plant. E. L., 220. 243 Porges. S. W., 293. 298 Porter. K . . 276,281,286. 2Y6 Posner. M . I . , 251, 254. 256,259. 265. 267. 269

Potter. M . c..25, 28. 30. 38 Poulson. C. L., 275. 281, 287, 294. 298 Pratt. W. W.. 113. 142 Pressley. M . , 206, 246 Presson. C . C.. 2. 3. 26. 38, 152. 160. 161, 164, 172, 191. 192, 200. 210. 215. 217, 218, 221.222. 223.224, 225, 231. 232, 233, 234. 238.239, 244. 246 Priefert. J. J., 273, 2YX Profitt. D. R.. 250, 268 Pufall. P. B.. 14, 3X. 237. 246

R kadin, N . . 107. I42 Radzisweska. B . , 106. 142 Ramsay, D. S.. 277. 278.296 Rand. Y.. 94, 114, 132. 139 Ratcliff. R.. 1.51, 200 Ratncr. H. H.. 90. 110. I I I. I42 Reese, H . W.. 99, 133. 143 Reeve. R . A . . 109. 142 Reiner. M . B.. 263, 269 Reiser. B. J . , 151, 19Y Reissland, N.. 290. 292. 294, 298 Renninger. K . A . , 106, 142 Renshaw. P.. 114, 115, 118, I42 Richards. H. C.. 273. 296 Richman. G . S . . 273. 296 Riegel. K. S . . 93. 142 Rieser. 1.. 207. 215. 246 Rivers. L., 120. 1-38 Robb, M . . 107. ISX Robinson, A. H . , 163. 200 Rodgon. M . M . . 277. 280.293. ?9X Rogoff, B.. 90.92.94,96, 101. 104, 105, 106, 107. 108. 109, 110. 112. 113. 118, 138. 13Y.

142. 152, I99 Romano, N . . 240. 244 Rook-Green. A . . 31. 3R

306

Author Index

Rosengren. K . S , , 107, 11 I , 137, I42 Rosenthdl, T. L . , 118, 142 Rosa. B. H . , 38 Ross, G.. 108, I44 Rosser, R . A . , 223, 224, 225, 226. 228. 23 1, 233. 244, 246 Roth, S. E , 238, 243 Rothbart, M., 265, 269 Roy, P., 114, 115, 140 Rozin, P.. 109, 142 Ruff, H., 250, 269 Rushton, J. P.,206, 246 Russel, M., 291, 297 Russell, J., 237, 246

S Salatas, H., 208, 232, 246 Sale, R. D., 163, 200 Samuels, S. J.. 262, 269 Sander, L. W., 93, 138 Saxe, G. B . , 90, 108, 142 Schachter, D., 224, 246 Schadler. M., 39, 152, 200 Schaeffer, B., 273, 297 Sckdller, J., 277, 292, 296 Schatzow, M. D., 222, 246 Schnell. G. A . . 148, 194, 200 Schnur, P., 257,258,269 Scholnick, E. K . , 152, I61 200 Schroeder, H. E., 297 Schubert. D. S. P.. 122, 143 Schubert, H. J. P., 122, 143 Schwartz, S . . 152, 161, 200 Scribner, S., 28, 36, 113, 142 Seitz, V., 272, 298 Selman, R. L., 134, 143 Shantz, C. U.. 42, 87,206, 222. 224, 230, 246 Shaw, R. E . . 14. 38,237,246 Shepard, R. N., 66,85, 87. 151,200, 217, 246 Sherman, J. A., 272,273,275, 287, 294, 295. 296 Shipman, V. C., 91. 140 Shipstead, S. G., 209, 210, 243 Shore, C., 294. 295 Short, K. R., 250,269 Sidman, M., 298 Siegel, A. W.. 9, 10.23,25, 39, 1.52, 200, 215, 243

Sigafoos, A . D., 276,281. 286, 290, 295 Silverman. I. W., 98, 143,224, 246 Simon, H. A , . 58. 86 Sims-Knight, J., 208. 209, 245 Skon, L., 114, 115. 143 Smith, C. D., 147, 200 Smothergill, D. W., 257, 259. 260. 261, 262, 263, 264, 268, 269 Snell, M. E., 273, 296 Snow, C. E., 84.87, 107, ion, 141 Snyder, L . , 294, 295 Sodian, B., 83, 8 7 Solla. J., 218, 243 Somerville, S. C.. 2. 38, 215. 241. 246, 247 Southworth, M., 148, 200 Southworth, S., 148. 200 Spelke. E., 152, 158, 159, 170, 199, 216, 244 Spencer, C.. 165, 167, 168, 169, 170, 172. 176, 198, 200 Spencer, W. A., 250, 270 Spelner, N. B., 250, 268 Spies, C., 217, 245 Sroufe. L. A., 103. 14.3 Stea, D., 164, 165. 172, 173, 190, 198, 201 Steinberg, Z., 114, 138 Steinman, W. M.. 273, 2Y8 Stern. D. N., 93, 140, 143 Stevenson, H. W., 92, 101, 143 Stevenson, M. B., 276. 277,280, 281, 282. 283,293,295 Stone, C., 114, 138 Strauss, M. S., 25,37, 250, 268 Strayer, E F., 134, 143 Streissguth. A. P., 91, 137 Striefel, J. A., 273, 298 Strommen, E. A , , 243 Sullivan, H. S., 10.5, 143 Super, C., 101, 143 Surber, C. F., 218,223, 228, 229, 243 Syc, S . , 114, 116. 143 Sylva, K., 110. 143 Symons, D., 284. 285, 297 Szeminska, A , , 245

T Taylor, M., 83, 87 Terrace, H. W.,273, 298 Thomas, H., 238, 247

Author Index

Thomas, M. A . , 250, 268 Thompson, R. F., 250. 270 Tinsley. V. S . . 95, 143, 206, 247 Todd, C. M . , 107, 138 Toupin. C., 19. 20. 27. 3 7 Treib, M . , 148, 201 Trudel, M . . 134, 143 Tulkin. S . R., 283, 284, 298 Tiriel, E.. 96, 114, 143 Tutton, A . , 284. 285, 297

U Underwood, B. J., 259, 270 Ungerer. J. A , . 25, 31. 39 Uttal, D H., 152, 175. 201 Uzgiris. 1. C., 277. 278. 279, 280, 281, 283. 284,285,294, 297. 298

V

Valentine. C. W., 274. 275, 298 Van Egeren. L. F.. 91, 137 Vaughn, B.. 208, 209. 24.5 Vega-Lahr, N., 276. 281, 286, 296 Verkozen, J., 209, 247 Vinter. A , . 298 Volterra, V.. 294, 295 VOSnkddoU. s.,39 Vurpillof, E . . 231, 247 Vygotsky. L. S., 2, 31, 39, 93. 95, 99, 101, 111. 127. 129. 143

W Wacha. T D , 283, 294. 298 Wagner. D A , 101, 143 Wagner, M E . 122. 143 Walker, L D , 221. 247 Warren. S F , 275.281.287, 298 Water\, H S , 95, 143. 206. 218, 223, 228. 229.230, 244, 247 Watson, J S , 222,224, 230, 246, 2Y6

307

Watson. M . W., 278,280. 282, 298 Waxler, C. 2.. 275, 286, 294, 298 Webb, N . , 132. 143 Wellman. A. M.. 152, 175, 201 Wellman, H. M., 42,46,47,48.49. 50. 56.80, 83. 86, 87, 111, 116. 143 Wendrich, K . A . , 279, 296 Wenters, J . , 240, 244 Wertsch, J. V., 92.93.94, 95, 107, 108, 142. 143. 144 Whalen, C., 273. 297 Willetts, E . . 223, 230, 242 William-Olsson. W., 165, 199 Wilson, D. T.. 58, 84, 8 7 Wimmer, H . , 33, 34,35, 38, 39. 83. 8 7 Winner, E., 3, 39 Witte, K. L., 251,268,270 Wohlwill, J. E , 147, 181, 195, 201 Wolcox. S . A , , 208,243 Wolf. D. P., 153, 159, 162, 163, 164. 175, 184, 193, 194, 196, 199, 200, 201 Wood. D. J., 90.94,95, 104, 107, 108, 109, 144 Wood. H. A , , 144 Woodson, R., 281,286,289,290,292. 294, 296 Wright. J. W., 109, 113. 139, 221, 243 Wynegar. L. T., 109, 144

Y Yaeger, J . , 95, 140 Yando, R . , 272, 2YX Yarrow. M R.. 215,286.294, 29X Yeates. K. O., 134. I43 Youniss, J . . 102. 10s.144,221,222,237,238. 242. 246

2

Zajonc, R. B.. 122, 144,267, 269 Zaporozhets, A. V.. 95, I44 Zelazo, P. R., 25,39 Zigler, E., 272, 298 Zimmerman, B. J., 99, 100, 115, 118, 142. 144

This Page Intentionally Left Blank

SUBJECT INDEX

A

developmental change, 124, 125, 129- 13I peer interactions, 113-115. 118, 119 perspectives, 93, 96, 97, 101 social agents, 104, 108 spatial perspective taking and, 205, 240, 241 display, 225-229 egocentrism, 207, 208 encoding. 216,217 mental rotation, 239, 240 operative development, 212 response mode, 231-234 rules for seeing, 210 tasks. 218,221,222,224, 225, 235, 236 Aggression, social influences on cognition and, 122 Alertness. stimulus repetition and, 251-253, 265,267,268 attention, 254-256 mechanisms, 259, 260 reading acquisition, 262-264 stimulus characteristics, 260-262 Alertness decrement, stimulus repetition and, 255, 256,265, 268 mechanisms, 261,262 reading acquisition, 264 Allocentric placement, spatial perspective taking and. 205,237,238 Analogical reasoning map concepts and, 177 representation and, 2, 12, 15-25 Analogous objects, representation and, 8- 11 Analogy mental phenomena and, 81 spatial perspective taking and, 228

Achievement, social influences on cognition and, 90 developmental change, 132 peer interactions, 116 perspectives, 101 social agents, 104, 122 Active participation, social influences on cognition and, 118 Affect, stimulus repetition and, 267, 268 Age imitation in infancy and developmental trends, 275-280 methods of evoking imitation, 285, 286 mother-infant interaction, 284 nonvisible actions, 288 trends, 293 variables, 281-283 map concepts and. 190 disciplinary traditions, 155. 156, 159, 170 Mapping Project at k n n State, 182, 187 symbolic representation, 194 mental phenomena and close imposters, 53-57 Laurendeau and Pinard. 79 mental images, 62-65 photographs, 72, 74-76 representation and, 33, 35, 36 analogical reasoning, 19.21-23 hidden object, 4 rapid changes, 8- I I spatial cognition, 14 symbolization, 32 social influences on cognition and, 91, 136

309

3 10

Subject Index

Arithmetic, mental phenomena and, 84 Associative learning, stimulus repetition and, 258 Asymmetric interactions, social influences on cognition and perspectives, 96, 102 social agents, 104, 105, 121, 123 Asymmetrical translation effect, representation and, 8- 10 Attention imitation in infancy and, 281, 282 map concepts and, 181 mental phenomena and, 59 spatial perspective taking and, 219 stimulus repetition and, 251, 254, 265 alertness decrement, 255, 256 orientation, 254, 255 reading acquisition, 262-264 Attributes, imitation in infancy and, 280 Autism, imitation in infancy and, 273 Automaticity theory, stimulus repetition and, 262,264

B Birth order, social influences on cognition and, 122

C Cartesian space, spatial perspective taking and, 214,216 Categories imitation in infancy and, 284 map concepts and, 190, 191 Mapping Project at F'enn State, 181, 185 symbolic representation, 194 mental phenomena and close imposters, 51,52, 54, 56. 57 Laurendeau and Pinard, 79,80 mental images, 64, 65 photographs, 74 representation and, 28 social influences on cognition and, 109 stimulus repetition and, 260. 263 Childhood realism, mental phenomena and, 42-45,56,57

Chunks, map concepts and, 165 Circular reactions, imitation in infancy and, 274 Close imposters, mental phenomena and, 49-57.68, 82 Cognition map concepts and conventional wisdom, 175 disciplinary traditions, 165 Mapping Project at k n n State, 178, 189, 190 symbolic representation, 195 social influences on, see Social influences on cognition spatial perspective taking and, 221,227,237, 240 Cognitive competence, social influences and, 110 Cognitive conflict, social influences and, 99 Cognitive development imitation in infancy and, 280 map concepts and, 180, 194 mental phenomena and, 85 representation and, 18, 28, 32, 36 social influences and. 90,91, 136 methodological issues, 133, 135 peer interactions, 112, 114. 120 perspectives, 95-98, 100, 101, 103 social agents, 103-11 I , 122, 123 stimulus repetition and, 250 Cognitive growth, social influences and, 90 peer interactions, 114 perspectives, 96.97, 100, 101 social agents, 108, I l l Cognitive processes imitation in infancy and, 272, 274, 279, 292 map concepts and, 155 mental phenomena and, 58 stimulus repetition and, 267, 268 Cognitive psychology, representation and, 2 Cognitive representations, mental phenomena and, 48 Cognitive science, mental phenomena and, 42, 85 Cognitive skills map concepts and, 147, 150, 177 representation and, 12, 15 social influences and, 135 methodological issues, 133, 134 peer interactions. 114. 117 perspectives, 94, 102

Subjecr Index

sibling interactions, 121 social agents, 104, 107, 110, 123 Cognitive structures imitation in infancy and, 293 social influences and. 96 Cognitive style. spatial perspective taking and, 218. 219 Cognitive tasks, spatial perspective taking and, 204,206 Coherence, map concepts and, 148 Collaboration, social influences on cognition and, 91 developmental change. 127 methodological issues, 135 peer interactions, 114-116, 119 perspectives, 97, 102 social agents, 104 theoretical issues, 132, 133 Competition. social influences on cognition and. 121. 122, I34 Componential level, map concepts and, 180 Comprehension map concepts and, 184, 191 stimulus repetition and, 262 Compromise, social influences on cognition and, 102, 105 Concrete operations, spatial perspective taking and, 210-212,240 Conditioned passivity. stimulus repetition and, 25 1 Conditioning, stimulus repetition and, 257. 258, 261 Conflict social influences on cognition and developmental change. 124, 126, 127, 129 peer interactions, 115, 116, 119. 120 perspectives, 97,99- 103 social agents. 105 spatial perspective taking and, 208. 220 Conformity, social influences on cognition and, 105 Conservation, stimulus repetition and, 250 Consistency, mental phenomena and, 46,47 Constraint, social influences on cognition and. 91. 136 developmental change, 124. 130 methodological issues, 133 peer interactions, 112 perspectives, 97,99 Constructivism, mental phenomena and, 44

311

Context map concepts and conventional wisdom, 176 developmental inquiry, 150 Mapping Project at Penn State, 183, 184 power, 147 symbolic representation, 194. 196 mental phenomena and, 82 representation and analogical reasoning, 15. 22 hidden object, 4 rapid changes, 5 symbolization, 31 social influences on cognition and, 90, 136 developmental change, 124, 128 peer interactions, 112, 115, 116 perspectives, 84.99 social agents, 103. 107, 109-112 spatial perspective taking and, 220. 235 Contingency, imitation in infancy and, 273, 286 Continuity, map concepts and, 148 Contradictory beliefs, representation and, 34 Convention, map concepts and, 183, 195 Conversations of gestures, 92, 93. 106 Coordinate system, spatial perspective taking and, 216 Correspondences, map concepts and. 177, 178, 182- 190 Cues imitation in infancy and, 280 map concepts and, 188 mental phenomena and, 84 representation and, 14 social influences on cognition and, 106. 109, 133 spatial perspective taking and, 221, 222, 237, 238 stimulus repetition and, 250

D Deferred imitation in infancy. 276, 279 Delay imitation In infancy and, 279 representation and, 4 stimulus repetition and, 265 Delayed testing, stimulus repetition and. 259, 260 Dialectical materialism, 93

312

Subject Index

Dimensionality, spatial perspective taking and, 227,235 Disequilibrium, social influences on cognition and, 115, 116, 127 Distance map concepts and, 152, 164, 174 spatial perspective taking and, 215 Distortion, map concepts and, 162, 163 Distraction social influences on cognition and, 125 spatial perspective taking and, 221 Dreams, mental phenomena and, 82 close imposters, 52 current studies, 48 Laurendeau and Pinard, 78, 81 menial entities, 45 Dual orientation, representation and, 31 Dual representation, 27, 28

E Egocentrism. spatial perspective taking and, 204-208,240 display, 229 response mode, 230,232,233 rules for seeing, 209. 210 tasks, 218, 220, 221, 223 Emotion, imitation in infancy and, 292 Enclosure, spatial perspective taking and, 212 Encoding spatial perspective taking and. 204, 2 12-2 I4 criticisms, 214-216 display, 226. 228, 229 model, 216, 217 response mode, 233,234 tasks, 220 stimulus repetition and, 251-253,265,267, 268 attention 254-257 mechanisms, 260 reading acquisition, 262-264 stimulus characteristics, 260-262 Epistemological realism, 42, 43, 83, 85 Euclidean space, spatial perspective taking and 204,236-238 encoding, 213,214,216 operative development, 211, 212 Extrapolation, map concepts and, 150, 162

Extroverts, social influences on cognition and, 132 Eye contact. social influences on cognition and. 106

F Facial expression imitation in infancy and, 281, 286, 289, 290 social influences on cognition and, 106, 109 spatial perspective taking and, 230 Facial gestures, imitation in infancy and, 274-276,284 Facilitation representation and, 18, 19, 21-23 social influences on cognition and, 90, 91, 135, 136 developmental change, 124, 127, 129 methodological issues, 133- 135 peer interactions, 115, 117, 120 perspectives, 97, 99, 101, 103 sibling interactions, 121- 123 social agents, 104, 107, 109, 110 theoretical issues. 132, 133 spatial perspective taking and, 219, 223 stimulus repetition and, 265, 267, 268 attention, 256 mechanisms, 257, 258 reading acquisition, 262,264 stimulus characteristics, 260-262 Fading, imitation in infancy and, 273 False belief attribution, representation and, 33-35 Familiarity imitation in infancy and, 277, 278 map concepts and, 154, 190 representation and rapid changes, 5 , 7, I I spatial cognition, 13, 14 social influences on cognition and, 113 spatial perspective taking and, 225 stimulus, see Stimulus repetition Fantasizing, mental phenomena and, 42 Feedback imitation in infancy and, 274, 287 social influences on cognition and, 121, 127 spatial perspective taking and, 224 Field independence, spatial perspective taking and, 218

Subject Index

Flexibility, representation and, 32.33. 36 hidden object. 4 symbolization, 28. 30 Forced choice. imitation in infancy and, 292 Formal operations, spatial perspective taking and, 210 Frame of reference map concepts and. 150 spatial perspective taking and, 234 Framework, spatial perspective taking and, 240 encoding, 216, 217 Euclidean space, 237, 238 mental rotation, 239 response mode, 233,234 tasks, 222.224,235,236 Free play imitation in infancy and, 279-281, 284 social influences on cognition and, 130 Friendship, social influences on cognition and. 119- 121 Frontedness, spatial perspective taking and. 226,227,229 Functional stimuli, stimulus repetition and. 259

G Gender, social influences on cognition and, 122 Generalization, social influences on cognition and. 109, 110, 127. 128. 130 Generalized imitation in infancy. 272. 287 Geometric correspondences, map concepts and, 177. 183-185, 187, 190 Geometry, spatial perspective taking and, 212 Gestural imitation in infancy developmental trends, 275-279 trends. 293. 294 variables, 282, 283 Goal structuring, social influences on cognition and, 111 Guillaume, imitation in infancy and, 274, 275

H Habituation imitation in infancy and, 286, 288 stimulus repetition and, 250, 251, 265 Harvard Project Zero, 159. 160

313

Hidden object mental phenomena and. 60-65 representation and, 2-4 analogical reasoning, 16, 20-23 rapid changes, 5-10 symbolization, 28-30 spatial perspective taking and, 209. 240 Home Stimulation Scale, imitation in infancy and, 283 Horizontality, spatial perspective taking and, 205.236, 238

I Iconicity, 183, 195 Idealism, mental phenomena and, 43 Imagination. mental phenomena and, 74, 79 Imitation, social influences on cognition and developmental change, 124 perspectives. 98-100, 103 social agents, 111 Imitation in infancy, 272-274 developmental trends, 280,281 deferred imitation, 279 gestural imitation, 275-279 early conceptualizations, 274, 275 methods of evoking imitation, 285-288 mother-infant interaction, 284, 285 nonvisible actions, 288-293 trends, 293-295 variables, 281, 283, 284 attention to model, 281, 282 latency, 282 object permanence, 282 upbringing, 282, 283 vocal imitation, 279, 280 Implicit knowledge, mental phenomena and, 77 Impulsivity, spatial perspective taking and, 218 Incomprehension. mental phenomena and, 79, 80 Infant imitation, see Imitation in infancy Inference map concepts and, 153, 162, 192 mental phenomena and, 42 representation and, 10, 15, 16, 24 spatial perspective taking and, 222. 227, 231 Inhelder, spatial perspective taking and, 204 display, 225, 227 egocentrism, 205

3 I4

Subjecr Index

encoding, 212-2 I6 mental rotation, 238, 239 operative development, 211, 212 response mode, 230,231,233, 234 rules for seeing, 210 tasks, 220 Inhibition representation and, 30, 32 social influences on cognition and, 109, 115, 123 stimulus repetition and, 267 age, 257, 258 mechanisms, 260 reading acquisition, 263, 264 Integral realism, mental phenomena and, 79, 80 Intellectual realism representation and, 33 spatial perspective taking and, 217, 218, 223 Intelligence, imitation in infancy and, 274 Intentionality, social influences on cognition and, 104 Interaction imitation in infancy and, 274 methods of evoking imitation, 286 mother-infant interaction, 284, 285 trends. 294 social influences on cognition and, 91, 135, 136 developmental change, 124, 128, 130, 131 methodological issues, 133-135 parent-child, 107- 112 peers, 112-121 perspectives, 93-98, 100- 102 siblings, 121, 122 social agents, 103- 107, 123 theoretical issues, 133 Interference, stimulus repetition and, 262, 263 Internalization imitation in infancy and, 275 social influences on cognition and, 136 developmental change, 128 perspectives, 93, 95, 102 social agents, 11 I Interpolation. map concepts and, 150, 162 Interposition, spatial perspective taking and, 223,230 Interpretation, map concepts and, 195 Introspection, mental phenomena and, 58, 59. 83.84

Introverts, social influences on cognition and, 132 IQ, social influences on cognition and, 122

L Landmarks, spatial perspective taking and. 240 encoding, 215-217 response mode, 232-234 tasks, 222,223,235 Language imitation in infancy and, 272 map concepts and, 157, 158, 166. 177 social influences on cognition and, 90.93 Latency, imitation in infancy and, 281, 282 Latent inhibition, stimulus repetition and. 257, 258 Laurendeau mental phenomena and, 56, 77-81 spatial perspective taking and, 225, 236, 237 Limited capacity, stimulus repetition and, 254 Linguistic interactions, social influences on cognition and, 92, 93 Linguistic skills, imitation in infancy and, 272

M Map concepts. 146, 147, 190-193, 196, 197 conventional wisdom, 171-176 developmental inquiry, 149, 150 disciplinary traditions, 150, 151 geographic and environmental work, 165- 171 psychological work, 151-164 map concepts and, 193-196 Mapping Project at Penn State child’s concept, 180- 182 correspondence, 182- 190 structure, 176- 179 power, 147- 149 Mapping, representation and, 26, 27 analogical reasoning, 16, 18, 19 spatial cognition, 13, 15 Mapping Project at Penn State child’s concept, 180-182 correspondences, 182- 190 structure, 176-179 Masking, stimulus repetition and, 258

Subject Index

Matched-dependent behavior, imitation in infancy and, 272,280,282, 294 Matching, imifation in infancy and developmental trends, 281 methods of evoking imitation. 287,288 mother-infant interaction, 284 nonvisible actions, 288-291 Memory imitation in infancy and, 279 map concepts and, 150, 151, 165 mental phenomena and. 42.45,82 representation and, 2, 3 analogical reasoning, 19. 24 hidden object, 3, 4 rapid changes, 5, 8 . 9 , I1 social influences on cognition and, 90, 108, 111, 116 spatial perspective taking and, 221, 227, 240 stimulus repetition and, 250, 259 Mental entities, see Mental phenomena Mental imagery, map concepts and, 151 Mental images, 57-67,82,84, 85 imitation in infancy and. 275 photographs, 67-77 Mental phenomena, 41-43.82-86 childhood realism, 43-45 current studies, 48.49 close impostors, 49-57 Laurendeau and Pinard, 77-91 mental images. 57-67 photographs, 67-77 mental entities. 45,46 prior studies. 46-48 Mental retardation, imitation in infancy and, 273 Mental rotation. spatial perspective taking and, 205,217,238-240 Metacognition mental phenomena and, 42.84 social influences on cognition and, 111, 123, 131 Metaphor representation and, 3, 18 stimulus repetition and, 250 Metarepresentation, 3 1 Mitigated realism, mental phenomena and, 79, 80 Model building, spatial perspective taking and, 233, 234. 236 Model placement, spatial perspective taking and. 236

315

Model rotation, spatial perspective taking and, 231, 232,236, 238 Modeling. imitation in infancy and, 273 developmental trends, 277-281 methods of evoking imitation, 285-288 mother-infant interaction, 284 nonvisible actions, 288-291 variables, 282, 283 Mother-infant interaction. imitation in infancy and, 284,285 Motivation social influences on cognition and, 90 developmental change, 125, 126 perspectives, 93.98 spatial perspective taking and, 226, 235 Motor habits. spatial perspective taking and, 207. 208

N Naturalism, spatial perspective taking and. 219, 220 Negative response bias, mental phenomena and, 59,60 Negotiation, social influences on cognition and peer interactions, 116. 119, 120 perspectives. 93, 102 social agents, 105, 106 Nominal realism, mental phenomena and, 67 Nominal stimuli, stimulus repetition and, 259 Novelty social influences on cognition and, 99, 102 stimulus repetition and, 250, 263, 265, 267 age. 257 attention, 254-256 mechanisms. 259-261

0 Object permanence, imitation in infancy and, 281.282 Object substitution, representation and, 31 Occlusion, spatial perspective taking and, 223, 235 Ontological dualism, 83. 85 Ontological realism, 83, 85 childhood realism, 43-45 close imposters, 57

316

Subject Index

current studies, 48 Laurendeau and Pinard, 81 mental images, 58.59 photographs, 68 Operant learning imitation in infancy and, 273 stimulus repetition and, 251 Operational thought, spatial perspective taking and, 239 Operative development, spatial perspective taking and, 210-212 Orientation spatial perspective taking and, 226, 227, 229 stimulus repetition and, 251, 254, 255 Orphanages, imitation in infancy and, 28 I , 283 Other regulation, social influences on cognition and. 95

P Parent-child interactions, social influences on cognition and, 107-112 Passivity, stimulus repetition and, 251 Pathway activation, stimulus repetition and, 256 Peer interactions, social influences on cognition and, 135, 136 developmental change, 124, 127, 128, 131 methodological issues, 135 perspectives, 97,98 social agents, 105-108. 110, 112-123 Permanence, stimulus repetition and, 250 Personality social influences on cognition and, 127-129, 131 spatial perspective taking and, 206 Phenominism. representation and, 33 Photographs mental phenomena and, 83 close imposters, 56 current studies, 67-77 Laurendeau and Pinard, 81 mental entities, 46 representation and, 28-30 Piaget imitation in infancy and early conceptualizations, 274, 275 methods of evoking imitation, 285 nonvisible actions, 288

trends, 293 variables, 282 map concepts and, 191 conventional wisdom, 173. 174 disciplinary traditions, 156, 158 Mapping Project at Penn State, 180 symbolic representation, 195 mental phenomena and, 82 childhood realism, 43 close imposters, 56, 57 current studies, 48 Laurendeau and Pinard, 78, 79, 81 mental images, 58 photographs, 67.68, 76 social influences on cognition and developmental change, 127 peer interactions, 114 perspectives, 96,97,99, 101 social agents, 104, I l l spatial perspective taking and, 203, 204, 240 display, 225, 229 egocentrism, 205-208 encoding, 212-216 mental rotation, 238, 239 operative development, 210-212 response mode, 230, 231, 233,234 rules for seeing, 210 tasks, 218-220, 224 Picture-selection task, spatial perspective taking and display, 228 Euclidean space, 238 response mode, 230-235 Pinard, spatial perspective taking and, 225. 236, 237 Place Perception Project, map concepts and, 165 Preexposure, stimulus repetition and, 257, 258 Preoperational stages, map concepts and, 180 Preoperational thought, spatial perspective taking and, 239,240 Primary representation, 31 Priming map concepts and, 151 stimulus repetition and, 256 Projective space map concepts and, 178, 191 spatial perspective taking and, 204, 236, 238 egocentrism, 207 encoding, 213 operative development, 211, 212

Subject index

response mode, 234 rules for seeing, 209 Prompting, representation and, 8, 19 Prototypes. stimulus repetition and, 259,260 Proximal development, social influences on cognition and developmental change, 129 peer interactions, 118 perspectives, 94, 99 theoretical issues, 132 Proximity. spatial perspective taking and, 212, 214. 217 Pseudo-imitation, 274 Publicness, mental phenomena and, 66.77

R Reaction time spatial perspective taking and, 240 stimulus repetition and, 251 age, 258 attention, 255 mechanisms, 259,260 reading acquisition, 263, 264 stimulus characteristics, 260. 261 Reading acquisition, stimulus repetition and, 251,262-264 Realism mental phenomena and, 42-45.83.85 close imposters, 56. 57 Laurendeau and Pinard, 79-81 photographs, 77 prior studies, 47 spatial perspective taking and, 217,218. 223 Reality testing, social influences on cognition and, 110 Recall, representation and, 19 Reciprocity, social influences on cognition and peer interactions, 120 perspectives, 92.98, 102 social agents, 104, 106 Reconstruction, representation and, 9 Reference, spatial perspective taking and, 237, 240 Referents, mental phenomena and, 45,46, 57, 67 Reflection mental phenomena and, 58.83 spatial perspective taking and, 226

317

Reflectivity, spatial perspective taking and, 218, 2 I9 Refusal, mental phenomena and, 79.80 Rehabituation, stimulus repetition and, 251 Reinforcement imitation in infancy and, 272, 273 methods, 285-287 trends, 294 social influences on cognition and developmental change, 125, 126, 129, 130 perspectives, 98 stimulus repetition and, 251 Relationship social influences on cognition and, 103, 104 spatial perspective taking and encoding, 215 Euclidean space, 237, 238 response mode, 231, 233,234 tasks, 224,235 Repetition imitation in infancy and, 281, 283,287 stimulus, see Stimulus repetition Representation, 2, 3, 35, 36 analogical reasoning, 15-25 hidden object, 3 . 4 map concepts and, 193 conventional wisdom, 174 Mapping Project at Penn State, 177, 181-183, 187, 189 mental phenomena and, 82 current studies, 48 Laurendeau and Pinard, 8 I mental entities, 45, 46 photographs, 67-69, 71, 74 prior studies, 46 multiple, 32-35 rapid changes, 4-12 spatial cognition, 13- 15 spatial perspective taking and, 205, 240 egocentrism. 207 Euclidean space, 237 response mode, 231,232,234, 235 tasks, 217-219, 236 symbolization, 25, 26 maps, 26.27 model as symbol, 27, 28 photographs, 28-30 symbolic play, 30-32 Representational correspondences, map concepts and, 190, 192

318

Subject Index

Mapping Project at Penn State, 177, 179. 180, 182, 184 symbolic representation, 196 Response mode, spatial perspective taking and, 217 Role play, imitation in infancy and, 287 Rotation, spatial perspective taking and, 238-240 display. 225, 228 response mode, 231, 232 tasks, 221, 236

S Scaffolding, social influences on cognition and, 107, 108, 117, 118 Schizophrenia, imitation in infancy and, 273 Self-regulation, social influences on cognition and developmental change, 124 perspectives, 94,95, 98 social agents, 111-113 Semantic encoding, stimulus repetition and, 264 Sensorimotor development imitation in infancy and, 288 map concepts and, 180 spatial perspective taking and, 210 Serial habituation, stimulus repetition and, 250 Shaping, imitation in infancy and, 273 Sheffield Research Program, map concepts and, 167, 168 Shielding. spatial perspective taking and, 221 Sibling interactions, cognition and, 135 developmental change, 124. 127, 128 social agents, 113, 120-122 Similarity mental phenomena and, 81 representation and analogical reasoning, 18-25 rapid changes, 9 spatial perspective taking and, 215 Smiling, imitation in infancy and, 284-286 Social agents, cognition and, 122, 123, 135, 136 cognitive development, 103-107 developmental change, 128 methodological issues, 134 parent-child interactions, 107- 112 peer interactions, 112-121

perspectives, 95, 97 sibling interactions, 121, 122 Social influences on cognition, 90-92, 135. 136 developmental change, 124-131 methodological issues, 133- 135 perspectives, 92, 101-103 contemporary, 101 social learning, 98-101 sociohistorical, 93-96 structural, 96-98 symbolic interaction, 92, 93 social agents cognitive development, 103-107 parent-child interaction, 107- 112 peer interactions, 112-121 sibling interactions, 121, 122 theoretical issues, 131-133 Social input, cognition and, 127-129, 131 Social learning, cognition and developmental change, 125 perspectives, 98-102 social agents, 111 Social skills cognition and, 91 methodological issues, 133, 134 peer interactions, 114, 117, 120 perspectives, 102 social agents, 104, 107, 123 imitation in infancy and, 272 Social status, cognition and, 123 Socialization cognition and, 104 mental phenomena and, 84 Society, cognition and, 105 Socioeconomic status imitation in infancy and, 281,283,284,287 mental phenomena and, 69 Space, map concepts and, 191- I93 conventional wisdom, 172-175 Mapping Project at Fknn State, 178-181, 183. 185- 189 Spacing, social influences on cognition and, 122 Spatial cognition, representation and, 2-4 analogical reasoning, 21, 23. 25 imitation, 13-15 rapid changes, 12 symbolization, 26 Spatial content, spatial perspective taking and, 212

Subject Index

Spatial development, spatial perspective taking and, 205 Spatial location, representation and, 4 Spatial perspective taking, 203-205. 240, 241 egocentrism, 205-208 encoding, 212-214 criticisms, 214-216 model, 216, 217 Euclidean space, 225-230.236-238 mental rotation, 238-240 operative development, 210-212 rules for seeing, 208-210 tasks, 217. 235, 236 array. 221 display, 225-230 movement, 222-224 naturalism, 219, 220 outside landmarks, 222. 223 response mode, 230-235 shielding of display, 221 subject, 217-219 training, 224, 225 Spatial relationship, map concepts and developmental inquiry, 150 disciplinary traditions. 151-153, 156-159, 161, 163, 164 power, 148 Speech, social influences on cognition and. 95 Stimulus familiarization. see Stimulus reperirion Stimulus repetition. 250-253. 265-268 age, 256-258 attention, 254 alertness decrement, 255, 256 orientation, 254, 255 mechanisms, 258, 259 delayed testing, 259 stimuli, 259, 260 reading acquisition. 262-264 stimulus characteristics. 260-262 Subjectivism. mental phenomena and. 57, 79, 80 Substitution, representation and, 31 Successive approximations. imitation in infancy and, 285 Surprise, spatial perspective taking and, 230 Symbolic interaction, social influences on cognition and. 92.93 Symbolic play, representation and, 30-32 Symbolic representation. 2, 3. 25, 26, 36 map concepts and. 147. 193-197

3 I9

conventional wisdom, 174 developmental inquiry, 149, 150 disciplinary traditions, 153, 159, 162, 163 Mapping Project at Penn State, 177. 180. 183. 184, 187, 189 power. 149 maps, 26,27 mental phenomena and, 48 model as symbol, 27, 28 photographs, 28-30 rapid changes. 12 symbolic play. 30-32 Symmetrical interaction, social influences on cognition and perspectives, 96.97 social agents, 104. 105, 123 Symmetry, spatial perspective taking and, 225, 226, 228, 229

T Topographic cues, spatial perspective taking and. 237 Topological cues. representation and, 14 Topological space, spatial perspective taking and. 204 encoding, 212-216 operative development, 211. 212 Touching order, spatial perspective taking and,

^." LIL

Training. spatial perspective taking and, 224, 225. 235 Transfer representation and. 15, 17-20, 22, 23 social influences on cognition and, I19 spatial perspective taking and, 224 Transformation mental phenomena and, 82, 83, 85 mental images. 59.61-63.67 photographs, 69-73 spatial perspective taking and display, 226-229 encoding, 217.219 mental rotation, 239 response mode, 231,233, 234 tasks, 221, 224 stimulus repetition and, 259, 260 Transitive reasoning, spatial perspective taking and. 212

320

Subject Index

Translation, representation and, 4, 9 , 10 Tutoring, social influences on cognition and, 114, 117, 132

U Upbringing, imitation in infancy and, 281 -283 Uzgiris-Hunt Scale, imitation in infancy and, 282-284

Vocal imitation in infancy, 279, 280, 284, 285

W Warning signal, stimulus repetition and, 265 attention, 254-256 reading acquisition, 263, 264 stimulus characteristics, 261

V Valentine. imitation in infancy and, 274, 275 Verticality, spatial perspective taking and, 205, 236,238

Zone of proximal development, social influences on cognition and, 94,99, 132, 139

Contents of Previous Volumes

Volume I

A Developmental Approach to Learning and Cognition

Responses of Infants and Children to Complex and Novel Stimulation

Evidence for a Hierarchical Arrangement of Leaning Processes

Eugene S . Gollin

Gordon N . Cantor

Sheldon H. White

Word Associations and Children's Verbal Behavior David S. Palermo Change in the Stature and Body Weight of North American Boys during the Last 80 Years

Selected Anatomic Variables Analyzed for lnterage Relationships of the Size-Size, Size-Gain. and Gain-Gain Varieties

Howard V. Meredith

Howard V. Meredith AUTHOR INDEX-SUBJECT

Discrimination Learning Set in Children

INDEX

Havne W,Reese Learning in the First Year of Life Lewis P. Lipsitr Some Methodological Contributions from a Functional Analysis of Child Development Sidney W Bijou and Daniel M. Baer The Hypothesis of Stimulus Interaction and an Explanation of Stimulus Compounding

Charles C . Spiker The Development of "Overconstancy" Perception

in Space

Infant Sucking Behavior and Its Modification

Herbert K a y The Study of Brain Electrical Activity in Infants Robert J. Ellingson Selective Auditory Attention in Children

Eleanor E . Maccohy

Joachim F. Wohlwill

Stimulus Definition and Choice

Miniature Experiments in the Discrimination Learning of Retardates Bet.

Volume 3

J . House and David Zeamun

Michael D. Zeiler Experimental Analysis of Inferential Behavior in Children

Tracy S. Kendler and Howard H. Kendler AUTHOR INDEX-SUBJECT

krceptual Integration in Children

INDEX

Herberr L. Pick, Jr., Anne D. Pick. und Robert E. KIein Coniponent Process Latencies in Reaction Times of Children and Adults

Volume 2

Ravmond H. Hohle The Paired-Associates Method in the Study of Contlict

Alfred Custuneda

AUTHOR INDEX-SUBJECT

INDEX

Transfer of Stimulus Pretraining to Motor PairedAssociate and Discrimination Learning Tasks

Joan H. Cantor Volume 4

The Role of the Distance Receptors in the Development of Social Responsiveness

Richard H. Walters und Ross D. Parke

Developmental Studies of Figurative Perception

David Elkind

Social Reinforcement of Children's Behavior

Harold W. Stevenson

The Relations of Shon-Term Memory lo Development and Intelligence

Delayed Reinforcement Effects Glenn Terrell

John M. Belmont and Earl C . Butterfield

321

Contents of Previous Volumes

322

Learning. Developmental Research. and Individual Differences

Superstitious Behavior in Children: An Experimental Analysis

Frances Degen Horowirz Psychophysiological Studies in Newborn Infants

S. J . Hurt. H. G.Lertard, and H. F.R. Prechrl Development of the Sensory Analyzers during Infancy Yvonne Bruckbill and Hiram E. Fifzgerald The Problem o f Imitation

Michael D. Zeiler Learning Strategies in Children from Different Socioeconomic Levels

Jean L. Bresnahan and Martin M . Shupiro Time and Change in the Development of the Individual and Society

Justin Arorifreed

AUTHOR INDEX-SUBJECT

Volume I

Klaus F.Riegel

INDEX

The Nature and Development of-Early Number Concepts

Rochel Gelman Learning and Adaptation in Infancy: A Comparison of Models

Volume 5

Arnold J. Sameroff

The Development of Human Fetal Activity and lation to Postnatal Behavior

If6

Re-

AUTHOR INDEX-SUBJECT

INDEX

Trvphena Humphrey Arousal Systems and Infant Heart Rate Responses

Frances K. Grahum and Jan C. Jackson Specific and Diversive Exploration

Corinne Hutf Developmental Studies of Mediated Memory

John H . Ffavefl Development and Choice Behavior in Probabilistic and Problem-Solving Tasks L. R. Goitlet and Kathryn S. Goodwiin AUTHOR INDEX-SUBJECT

INDEX

Volume 8 Elaboration and Learning in Childhood and Adolescence

William D.Rohwer, Jr. Exploratory Behavior and Human Development

Jum C. Nunnally and L. Charles Lemond Operant Conditioning of Infant Behavior: A Review

Robert C. Hulsebus Birth Order and Parental Experience in Monkeys and Man

G. Mitchell and L. Schroers Fear of the Stranger: A Critical Examination

Harriet L. Rheingold and Carol 0. Eckerman Applications of Hull-Spence Theory to the Transfer of Discrimination Learning in Children

Volume 6 Incentives and Learning in Children

Charles C. Spiker and Joan H. Cantor

Sam L. Witryol AUTHOR INDEX-SUBJECT

Habituation in the Human Infant

INDEX

Wendell E. Jejjrey arid Leslie B. Cohen Application of Hull-Spence Theory to the Discrimination Learning of Children

Volume 9

Charles C . Spiker Growth in Body Size: A Compendium of Finding5 o n Contemporary Children Living in Different Parts of the World

Howard V. Meredirh Imitation and Language Development James A . Sherman Conditional Responding as a Paradigm for Observational. Imitative Learning and Vicarious-Reinforcenient

Jacob L. Gewirtz AUTHOR INDEX-SUBJECT

INDEX

Children's Discrimination Learning Based on Identity or Difference Berry J. House. Ann L. Bron'n.

and Marcia S.Scon Two Aspects of Experience in Ontogeny: Development and Learning Hans G . Furth The Effects of Contextual Changes and Degree of Component Mastery on Transfer of Training Joseph C . Campione and Ann L. Brown

Contents of Previous Volumes

Psychophysiological Functioning, Arousal, Attention. and Learning during the First Year of Life

Richard Hirschman and Edn'ard S. Karkin Self-Reinforcement Process in Children

John C. Masters and Janice R. Mokros AUTHOR INDEX-SUBJECT

INDEX

Volume 10 Current Trends in Developmental Psychology Boyd R. McCundless and Mary Fulcher Geis The Development of Spatial Representations of LargeScale Environments Alexander W. Siege1 and Sheldon H. Whire Cognitive Perspectives on the Development of Memory

John W. Hugen, Roben H. Jongeicvird. Jr., and Robert V. Kuil. Jr.

323

Theory and Method in Life-Span Developmental Psychology: Implications for Child Development Alerhti Huston-Sreiti and Paul B. Balres The Development of Memory: Life-Span Perspectives

Hayne W. Reese Cognitive Changes during the Adult Years: Implications for Developmental Theory and Research Nancy W. Denney find John C. Wrighr Social Cognition and Life-Span Approaches lo the Study of Child Development Michael J . Chandler Life-Span Development of the Theory of Oneself Implications for Child Development Orville G. Brim. Jr. Implications of Life-Span Developmental Psychology for Childhood Education Leo Montadu and Sigrun-Heide Filipp AUTHOR INDEX-SUBJECT

INDEX

The Development of Memory: Knowing. Knowmg About Knowing. and Knowing How to Know

Ann L. Brown Developmental Trends in Visual Scanning

Mary Carol Day The Development of Selective Attention: From Perccptual Exploration to Logical Search John C. Wrighr and Alice G . Vliersrra AUTHOR INDEX-SUBJECT

INDEX

Volume U Research between 1960 and 1970 on the Standing Height of Young Children in Different Parts of the World Howard V. Meredirh The Representation of Children's Knowledge D a d Kfuhrand Roherr S.Siegler Chromatic Vision in Infancy

Marc H. Bornstein Developmental Memory Theories: Baldwin and Piaget

Bruce M. Ro.rs and Srephen M. Kersr Volume I1 The Hyperactive Child: Charactenstics. Treatment, and Evaluation of Research Design Gladys B. Barley and Judith M. LeBlanc Peripheral and Neurochemical Parallels of Psychopathology: A Psychophysiological Model Relating Autonomic Imbalance to Hyperactivity. Psychopathy. and Autism

Child Discipline and the Pursuit of Self An Historical Interpretation

Howard Gadlin Development of Time Concepts in Children

Willium J. Friedman AUTHOR INDEX-SUBJECT

INDEX

Srephen W Porges Constructing Cognitive Operations Linguistically

Harry Beilin operant Acquisition of' Social Behaviors in Infancy: Basic Problems and Constraits W. Sruarr Millar Mother-Infant Interaction and Its Study

Jacob L. Genwtz and Elizabeth F. Boyd Symposium on Implications of Life-Span Developmental Psychology for Child Development: Introductory Remarks Paul B. Bulres

Volume I3 Coding of Spatial and Temporal Information in Episodic Memory

Daniel B. Berch A Developmental Model of Human Learning

B a r p Gholson and Harry Beilin The Development of Discrimination Learning: A Levels-of-Functioning Explanation Tracy S . Kendler

324

Contents of Previous Volumes

The Kendler Levels-of-Functioning Theory: Comments and an Alternative Schema

Charles C. Spiker and Joan H. Cantor Commentary on Kendler’s Paper: An Alternative Perspecti ve

Children’s Clinical Syndromes and Generalized Expectations of Conuol

Fred Rothbaum AUTHOR INDEX-SUBJECT

INDEX

Barry Gholson and Therese Schuepjer Reply to Commentaries Tracy S. Kendler On the Development of Speech Perception: Mechanisms and Analogies

Peter D. Eimus and Vivien C. Tartter The Economics of Infancy: A Review of Conjugate Reinforcement

Carolyn Kent Rovee-Collier and Marry J. Gekoski

The History of the Boyd R. McCandless Young Scientist Awards: The First Recipients

David Palermo Social Bases of Language Development: A Reassessment

Human Facial Expressions in Response to Taste and Smell Stimulation

Jacob E. Steiner AUTHOR INDEX-SUBJECT

Volume 16

Elizaberh Bates. Inge Bretherton, Marjorie Eeeghiy-Smith, and Sandra MrNew Rrceptual Anisotrophies in Infancy: Ontogenetic Origins and Implications of Inequalities in Spatial Vision

INDEX

Marc H. Bornstein Concept Development

Martha J. Farah and Stephen M. Kosslyn Production and Rrception of Facial Expressions in Infancy and Early Childhood

Volume 14 Development of Visual Memory in Infants John S. Werner and Marion Perlmutrer Sibship-Constellation Effects on Psychosocial Development, Creativity, and Health

Mazie Earle Wagner,Herman J . P. Schubert. and Daniel S. P. Schubert The Development of Understanding of the Spatial Terms Front and Back Lauren Julius Harris and Ellen A . Strommen The Organization and Control of Infant Sucklng

Tiffany M . Field and Tedru A. Walden Individual Differences in Infant Sociability: Their Origins and Implications for Cognitive Development

Michael E. Lamb The Development of Numerical Understandings Robert S. SiegIer and Mitchell Robinson AUTHOR INDEX-SUBJECT

INDEX

C. K. Crook Neurological Plasticity, Recovery from Brain Insult, and Child Development

Ian Sr. James-Roberts AUTHOR INDEX-SUBJECT

Volume 17 The Development of Problem-Solving Strategies

INDEX

Deanna Kuhn and Erin Phelps Information Processing and Cognitive Development

Robert Kail and Jeffrey Bisanz Volume 15 Visual Development in Ontogenesis: S o m e Reevaluations

Juri Allik and Juan Valsiner Binocular Vision in Infants: A Review and a Theoretical Framework

Richard N.Aslin and Susan T. Dumuis Validating Theories of Intelligence Earl C . Butterfield, Dennis Siladi,

and John M. Belmont

Research between 1950 and I980 on Urban-Rural Differences in Body Size and Growth Rate of Children and Youths

Howard V. Meredith Word Meaning Acquisition in Young Children: A Review of Theory and Research

Pamela Blewitt Language Play and Language Acquisition

Stan A. Kuczaj II The Child Study Movement: Early Growth and Development of the Symbolized Child

Alexander W. Siege1 and Sheldon H. White

Cognitive Differentiation and Developmental Learning

William Fowler

AUTHOR INDEX-SUBJECT

INDEX

Contenis of Previous Volumes

Volume 18 The Development of Verbal Communicative Skills in Children

Constance R. Schmidr and Scotr G . Paris Auditory Feedback and Speech Development

Gerald M. Siegel, Herbert L. Pick. Jr., and Sharon R. Garber Body Size of Infants and Children around the World in Relation to Socioeconomic Status

325

Effects of Sibling Spacing on Intelligence, Interfamilial Relations, Psychosocial Characteristics, and Mental and Physical Health

Mazie Earle Wagner,Herman J. P. Schuberr. and Daniel S. P. Schuberr Infant Visual Preferences: A Review and New Theoretical Treatment Marrin S. Banks and Arthur P. Ginsburg AUTHOR INDEX-SUBJECT

INDEX

Howard V. Meredith Human Sexual Dimorphism: Its Cost and Benefit James L. Moslev and Eileen A. Sran Symposium on Research Programs: Rational Alternatives to Kuhn’s Analysis of Scientific Progress-htroductory Remarks

Hayne W. Reese. Chairman World Views and Their Influence on Psychological Theory and Research: Kuhn-Lakatos-Laudan

WillisF. Overron The History of the Psychology of Learning as a Rational Process: Lakatos versus Kuhn

Perer Barker and Barrv Gholson Functionalist and Structuralist Research Programs in Developmental Psychology: Incommensurability or Synthesis?

Harry Beilin In Defense of Kuhn: A Discussion of His Detractors

David S. Palermo Comments on Beilin’s Epistemology and Palermo’s Defense of Kuhn

Volume 20 Variation in Body Stockiness among and within Ethnic Groups at Ages from Birth to Adulthood

Howard V. Meredirh The Development of Conditional Reasoning: An Iffy Proposition

David P. O’Brien Content Knowledge. Its Role, Representation, and Restructuring in Memory Development

Michelene T. H. Chi and Stephen J . Cecr Descnptions: A Model of Nonstrategic Memory Development

Brian P. Ackerman Reactivation of Infant Memory: Implications for Cognitive Development

Carolyn Rovee-Collier and Harlene Hayfie Gender Segregation in Childhood

WillisF. Owrron From Kuhn to Lakatos to Laudan

Perer Barker and Bar? Gholson Ovenon’s and Palermo‘s Relativism: One Step Forward, Two Steps Back

Eleanor E. Maccoby and Carol Nagv Jacklin Piaget, Attentional Capacity, and the Functional Implications of Formal Structure

Michael Chapman

Harry Beilin INDEX AUTHOR INDEX-SUBJECT

INDEX

Volume 21 Volume 19 Response to Novelty: Continuity versus Discontinuity in the Developmental Course of Intelligence Cynthio A. Berg and Roberr J. Sfernberg Metaphoric Competence in Cognitive and Language Development Marc Marschark orid Lvnn No11 The Concept of Dimensions in Developmental Research

Stuart I. Offenbach and Francine C. Blumberg Effects of the Knowledge Base on Children’s Memory Strategies

Peter A. Ornsrein and Mary J. Nous

Social Development in Infancy: A 25-Year Perspective Ross D. Parke On the Uses of the Concept of Normality in Developmental Biology and Psychology Eugene S. Gollin. Gary Sruhl. and Elyse Morgun Cognitive Psychology: Mentalistic or Behavioristic?

Charles C . Spiker Some Current Issues in Children’s Selective Attention

Bery J . House Children’s Learning Revisited: The Contemporary Scope of the Modified Spence Discrimination Theory

Joan H. Cantor and Charles C . Spiker

326

Contents of Previous Volumes

Discrimination Learning Set in Children Hayne W Reese A Developmental Analysis of Rule-Following Henq C . Riegler and Donald M . Boer Psychological Linguistics: Implications for a Theory of Initial Development and a Method for Research Sidney w. Rijou Psychic Conflict and Moral Development Gordon N . Cunror and David A . Purton

Knowledge and the Child’s Developing Theory of the World

David S. Palermo Childhood Events Recalled by Children and Adults

David B. Pilfemer and Sheldon H . White INDEX